681
AndroGuider - One Stop For The Techy You! Like us on Facebook: https://www.facebook.com/androidguider/ Follow us on Twitter: https://twitter.com/AndroGuider Google Plus page: https://plus.google.com/+AndroGuider
e industry fiercely opposed - but it gives the state clear enforcement tools and gives first responders a seat at the table in future permit reviews. How This Hits Waymo, Cruise, And Zoox
For Waymo, the market leader with more than 100,000 paid rides per week in San Francisco, Los Angeles, and the Bay Area, the impact is operational rather than existential. The company already has a first-responder hotline and training program with local fire departments, and it says its latest software is better at recognizing emergency vehicles. Still, it will need to expand its fleet response teams, upgrade its remote assistance staffing to meet the 30-second rule, and overhaul its incident reporting systems to meet the 24-hour deadline.
Cruise faces higher stakes. The GM-backed company is in the middle of a cautious comeback in California after its statewide suspension in 2023 following a pedestrian dragging incident in San Francisco. With plans to relaunch paid service in Phoenix, Houston, and parts of the Bay Area, any permit suspension under the new law would be a major setback. Cruise has publicly supported the bill and says it is building new emergency pull-over and two-way voice features into its vehicles.
Amazon's Zoox, which just began public operations in Las Vegas and Foster City with its purpose-built, steering-wheel-free shuttle, and smaller players like May Mobility and Motional, will also be covered. Industry groups warn compliance will raise costs, especially the requirements for continuous human monitoring, data storage, and city-by-city coordination drills with police and fire agencies. What Happens Next
The law takes effect January 1, 2027, giving the DMV three months to finalize the fine schedule, reporting portal, and technical standards for emergency detection.
Expect a flurry of activity before then. The DMV is required to hold public workshops with first responders and AV companies this fall, and Waymo and Cruise have already pledged to participate in joint training exercises in San Francisco and Santa Monica.
Civil liberties and labor groups say they will be watching enforcement closely, pushing for public disclosure of all interference reports. The AV industry, meanwhile, is lobbying for a clear appeals process to avoid fines for edge cases where yielding was technically impossible.
One thing is clear: as robotaxis go from novelty to everyday infrastructure in California, the era of a driverless car blocking a fire truck with no consequences is over.
or 90 days can cost a dealer well over $1,000 in holding costs while losing value. Rented at $800 to $1,100 per month through MyMonthlyCar, that same SUV can cover its flooring costs, generate gross profit, and stay freshly maintained with regular use and servicing.
It also opens a new customer funnel. Monthly renters are effectively extended test drivers. Data from car subscription pilots in Europe and the U.S. shows a significant share of subscribers convert to buyers of the same brand. For dealers, a renter today could be a buyer in three months — without any additional customer acquisition cost. The Bigger Bet on the Future of Car Access
MyMonthlyCar is riding a broader shift: from car ownership to car access. Younger drivers, gig workers, relocators, international professionals, and EV-curious shoppers increasingly prefer flexibility over a five-year loan. Automakers have experimented with subscriptions before, but most programs were too expensive and too centralized to scale.
Dobrianskyi is betting the future looks more decentralized — powered not by automakers or rental giants, but by the 18,000-plus franchised dealerships already embedded in every American community. If he is right, the local dealership lot could evolve into a neighborhood mobility hub: part showroom, part rental fleet, part subscription service.
Challenges remain, from insurance regulation across states to fleet management logistics and consumer education. But as MyMonthlyCar heads to the Disrupt stage in San Francisco, its message to dealers is clear: your next profit center is already parked outside.
position. Their job requires candid documentation of model risks, failures, and potentially dangerous capabilities — findings that can be uncomfortable for a company racing to ship. Some researchers privately worry that aggressive enforcement of confidentiality rules could chill legitimate internal dissent and make it harder to raise safety concerns.
Others argue strict controls are unavoidable. As models become more powerful, detailed knowledge of their weaknesses, training methods, and safety bypasses becomes a security risk in itself. Mishandling that information, even unintentionally, could expose the company to leaks, regulatory fallout, or misuse.
Former OpenAI staffers and outside experts have already begun debating the case online, with some calling for more transparency about what lines were crossed, and others defending OpenAI's right to enforce basic data hygiene. WHAT IT MEANS FOR AI SAFETY AND TRUST
The incident highlights a central tension defining OpenAI in 2026: can it move fast and stay trustworthy?
For regulators in Washington, London, and Brussels, the episode will likely reinforce demands for formalized safety governance, auditable handling of frontier model evaluations, and clearer whistleblower protections that distinguish between harmful leaks and good-faith safety disclosures.
For enterprise customers and partners, the firings may actually be reassuring — evidence that OpenAI monitors and enforces its own safeguards. In an era where AI models power critical business infrastructure, customers want to know proprietary data and pre-release safety findings won't be carelessly shared.
For the public, however, the opacity surrounding the case cuts both ways. OpenAI asks for trust that its models are safe and aligned, but key details about how safety decisions are made, who makes them, and what happens when internal rules are broken remain behind closed doors. WHAT TO WATCH NEXT
Several key questions remain unanswered.
First, will OpenAI disclose more about the nature of the mishandled information and whether any external parties were affected. Second, whether the three former researchers will speak out publicly or challenge the company's account. Third, how this will affect morale and retention within OpenAI's remaining safety teams, which have already seen significant turnover.
The company is expected to brief employees internally in the coming days and may update its internal data-handling policies further. Observers will also be watching for any response from the nonprofit board that oversees OpenAI, as well as from lawmakers focused on AI oversight.
One thing is clear: as the race toward more capable AI accelerates, the battle over who controls sensitive safety knowledge — and how carefully it is guarded — is becoming just as important as the technology itself.
Station, Fervo demonstrated commercial flow rates of more than 10 megawatts per well pair in testing, far above Department of Energy benchmarks from just a few years ago. Independent validation of its earlier 3.5-megawatt Project Red pilot in Nevada for Google in 2023 gave utilities confidence to sign on. Why Utilities Are Buying In Fast
Cape Station is already largely sold out. Southern California Edison has contracted for 320 megawatts, and Google has contracted for additional carbon-free power to run data centers, part of a broader push for 24/7 clean power for AI loads.
For utilities, the appeal is speed plus certainty. Solar and batteries are fast but intermittent. Nuclear is firm but can take a decade or more. Fervo is pitching geothermal as fast AND firm — new capacity in about two years with a 30-plus year plant life and stable pricing.
That timing matters as U.S. electricity demand surges for the first time in two decades, driven by data centers, manufacturing, and electrification. Grid operators in the West, where Cape will feed into the regional system, are urgently seeking nighttime, winter-proof power to complement massive solar buildouts. What's Next: Even Faster Grid Connections
Fervo says Phase 1 was the slow one. Phase 2 at Cape, adding another 400 megawatts, is already being drilled with the same pads, permits, transmission corridor, and plant design.
The company expects three big accelerators:
1. Learning curve drilling: With crews, geology, and bit designs now proven at Cape, Fervo targets days-per-well closer to shale norms, cutting months off each pad.
2. Standardized plant blocks: Phase 2 will replicate the Phase 1 power island design, allowing pre-ordered turbines, cooling, and electrical gear instead of custom engineering.
3. Ready grid access: The substation and 138-kV transmission tie-line built for Phase 1 were oversized for the full 500-megawatt project, meaning future electrons can flow with minimal new interconnection study delays.
Fervo is targeting full 500-megawatt operation at Cape by 2028, and is advancing a pipeline of over 1 gigawatt in Utah, Nevada, and California. Backed by more than $800 million in funding and a $148 million Department of Energy grant for EGS innovation, the company argues costs will fall with each phase, following a similar path to shale and solar. The Road To Scale
Challenges remain. EGS still requires large upfront drilling capital, careful water management in arid Utah, and long-term proof that artificially created reservoirs maintain heat and flow for decades. Seismicity monitoring and community engagement will stay under close scrutiny as operations scale.
But completing the world's first large-scale enhanced geothermal plant in 23 months reframes what's possible. If Fervo can repeat and shorten that timeline, geothermal shifts from a niche, slow resource to a deployable, repeatable clean power platform — one that can be built at the speed of the AI boom it aims to power.
comes plausible within the next decade.
In other words: no Starship, no space data centers. How Orbital Computing Would Actually Work
Google is not proposing to replace Earth data centers entirely.
The Suncatcher architecture envisions tightly networked satellite clusters connected by free-space optical links — essentially laser internet in space capable of terabits per second. Each satellite would combine TPUs, solar arrays, and radiators for heat rejection, flying in formation to act as a distributed supercomputer.
Workloads best suited for orbit would be highly parallel, energy-intensive, and fault-tolerant, like AI training, video transcoding, and large-scale inference. Latency-sensitive tasks and long-term data storage would stay on the ground.
Crucially, Google plans to use the vacuum of space itself for cooling. Without air, heat must be radiated away, which is both an advantage and a major engineering challenge. The prototypes now in orbit will test new radiator designs and chip layouts to see if TPUs can run at full power without overheating. The Massive Challenges Still in the Way
Despite the excitement, Google is candid that orbital data centers are far from inevitable.
Radiation is the first hurdle. High-energy particles can flip bits, degrade chips, and shorten lifespans from 7 years to months. Google's TPUs were never designed with radiation hardening, so the current mission will reveal whether software-level error correction is enough.
Thermal management is second. In direct sunlight, satellites swing between extreme heat and cold. Dissipating kilowatts of AI heat without fans or water cooling requires huge, lightweight radiators that have never been flown at this scale.
Then there is networking, collision risk, and maintenance. A gigawatt fleet would mean tens of thousands of satellites, dramatically increasing orbital debris concerns. Unlike a ground data center, you cannot send a technician to swap a failed TPU. Google will need autonomous rendezvous, robotic servicing, or simply accept high attrition rates.
Finally, cost and regulation remain open questions. Even with 1,800 Starship launches, the capital expenditure would run into tens of billions of dollars, and international rules on spectrum, orbital slots, and light pollution are still catching up. What Comes Next
Google says the next milestone is a scaled demonstration in 2027: a small constellation of solar-powered TPU satellites operating as a single system with space-to-space laser links and direct downlink to Earth.
If that works, the company will evaluate a 10-megawatt operational prototype — still tiny compared to a 1-gigawatt terrestrial campus, but large enough to prove the economics.
Competitors are watching closely. SpaceX, Amazon, and startups like Starcloud and Axiom Space have all floated similar space-compute concepts, and the Pentagon has expressed interest in resilient orbital AI.
For now, Google frames Suncatcher as research, not a product roadmap. But the message is clear: as AI hits the limits of Earth's energy system, the next cloud region may not be in Iowa or Belgium — it may be 500 kilometers straight up.
urated product picks based on your past chats, saved Favorites, and stated preferences like size, budget, and style. If you've been shopping for minimalist home decor and trail running shoes, For You will adapt accordingly. OpenAI says you can edit or clear these preferences and shopping memories at any time. Why This Matters: The Future of AI-Powered Shopping
This launch is much bigger than a fun photo filter. It's OpenAI's clearest move yet to own the entire shopping journey - from discovery to decision to checkout.
ChatGPT already became one of the web's top product discovery tools after OpenAI added shopping results last year. Now with visual try-on, persistent memory, and Instant Checkout via its Agentic Commerce Protocol, it can compete directly with Google, Amazon, and TikTok Shop.
For merchants, the stakes are huge. OpenAI says product listings are pulled from third-party feeds including Shopify, Etsy, and other catalog partners, with no pay-to-rank placement. Brands that provide high-quality images, accurate sizing data, and 3D-ready assets will be more likely to be try-on compatible and recommended.
For shoppers, it points to a future where you no longer scroll through dozens of tabs and reviews. Instead, you describe your need, see it on yourself, save it for later, and let AI watch the price.
OpenAI plans to expand try-on to more categories like beauty, eyewear prescriptions, and home furniture placement, plus add multi-person try-ons and video previews. As one executive put it during the announcement: search helps you find clothes, but ChatGPT wants to help you know if you'll actually wear them.
ts and probabilities, while AI gives direct answers with bullet-point clarity. For a president who values gut instinct and speed, that style is persuasive, even if the underlying data is drawn from public information, not classified intelligence. Inside the White House's Growing AI Habit
The Maduro episode is just the most dramatic example of AI creeping into the Situation Room. Under Trump's second term, the administration has pushed an AI-first governing doctrine, integrating commercial models from xAI, OpenAI, and Palantir into federal agencies for everything from border vetting to regulatory rollbacks.
Musk's dual role as both tech provider and presidential confidant has blurred lines. Grok is now embedded in X, federal employee tools, and reportedly in a custom White House deployment dubbed Grok for Government. Ethics experts warn this creates a feedback loop where the President gets strategic advice from a product owned by his top donor and advisor.
Supporters argue it's modernization. Critics call it outsourcing war powers to Silicon Valley with zero oversight, no audit trail, and no understanding of how the model reaches conclusions. Experts Warn: AI Should Not Call the Shots
National security and AI safety experts are alarmed, regardless of whether Grok's advice was good this time. Large language models hallucinate, lack access to real-time classified intelligence, and are trained to be agreeable — a dangerous combination for military planning.
Pentagon veterans note Grok has no clearance, no knowledge of troop positions, and no accountability. It cannot be court-martialed if it's wrong. Relying on it, even informally, risks leaking sensitive operational details into commercial servers and normalizing chatbot warfare.
Lawmakers from both parties have since called for hearings on AI use in lethal decision-making, and for a clear ban on using public chatbots for operational planning. The White House says all proper protocols were followed and that Grok was used for open-source context only, not for classified targeting. What This Reveals About the Future of War
Whether Grok tipped the scales or simply confirmed what Trump already wanted to do, the story marks a turning point. For the first time, a U.S. president is reported to have consulted a commercial AI in the decision chain for capturing a foreign head of state.
It signals a future where wars may be debated not just by generals and diplomats, but by prompts and chat completions. As AI models become more embedded in government, the questions will only get harder: Who audits the advisor, who owns its data, and what happens when the chatbot is wrong?
For now, the Maduro capture is being hailed by the administration as a triumph of decisive leadership. But in Silicon Valley and Washington, the quieter takeaway is stark — Grok is no longer just answering questions on X. It's in the room where it happens.
ays to minutes. Meanwhile, there is a global shortfall of nearly 5 million cybersecurity professionals.
Venture firms see agent swarms as the only way to close that gap. Static rules and single-agent assistants can't keep up with attacker swarms that probe thousands of vectors at once. Only a defensive swarm can match scale with scale.
Mandia brings something few AI founders can: two decades of frontline breach credibility from the Sony hack to SolarWinds to Colonial Pipeline. For boards terrified of AI breaches, that trust is the moat. What It Means For The Future Of Cyber Defense
If Armadin delivers, it could kill the security model as we know it. The era of human-led alert triage, quarterly red teaming, and 100-plus tool sprawl would give way to lean, autonomous platforms where testing and response happen continuously.
Rivals won't sit still. CrowdStrike, Palo Alto Networks, Google Mandiant, and a wave of AI-native startups like XBOW and Seven AI are all racing toward agentic security. Expect rapid consolidation, talent wars for AI red-team researchers, and fierce debate over guardrails: who is liable when an autonomous agent quarantines a hospital network or blocks a trading desk?
Mandia acknowledges the stakes. Armadin is launching with a containment-first architecture, human override controls, full audit trails, and FedRAMP and SOC 2 Type II compliance in progress.
But his message to CISOs is blunt: attackers already have swarms. Defenders now need their own.
PO-era matter behind it so it can focus on customers, drivers, and long-term growth. What It Means for Drivers
For the roughly 800,000+ drivers who use Lyft regularly in the U.S. and Canada, the short answer is: nothing changes.
No back pay, no reclassification, no new benefits flow from this shareholder settlement. The money goes to eligible investors, minus attorneys’ fees typically around 25-30%, not to drivers.
Driver advocates say that is exactly the problem. While Lyft will pay more than a quarter-billion dollars to investors over statements about driver labor law, drivers themselves continue to fight over take rates, deactivations, and whether Prop 22’s earnings floor — calculated on engaged time, not waiting time — truly amounts to a living wage.
Labor groups including Rideshare Drivers United and the Service Employees International Union said this week the settlement shows how much value is tied to keeping labor costs low, and renewed calls for collective bargaining rights and federal protections regardless of contractor status.
Lyft, for its part, points to recent concessions: higher minimums in several cities, more transparent fare breakdowns, appeal processes for deactivations, and the health stipends required under Prop 22. It insists most drivers prefer contractor flexibility over set schedules. The Bigger Picture for the Gig Economy
Lyft’s $272.5 million deal is less about 2026 and more about 2019 — a time when the entire gig model was legally fragile. Six years, one ballot measure, and one California Supreme Court decision later, that model has proven remarkably resilient.
Uber settled a similar IPO-era shareholder case for $200 million in 2024, and DoorDash has faced its own disclosure suits. Together, the payouts show investors are willing to punish tech companies retroactively for regulatory risk, even if the companies ultimately win the regulatory war.
The next frontier is not AB5, but automation and pay transparency. As robotaxis expand in Los Angeles, San Francisco, Austin and Phoenix, both Lyft and Uber are telling investors human drivers will remain central for years — a claim that could itself become the subject of future disclosure lawsuits if it proves wrong.
For now, Lyft has bought peace on its past. Whether it can avoid a repeat with its future remains the $272.5 million question.
AndroGuider | One Stop For The Techy You!
Lyft Pays $272.5M to Settle Driver Classification Lawsuit
https://ai4chat-files.s3.amazonaws.com/images/image_1790900495191.jpg TL;DR
* Lyft has agreed to pay $272.5 million to settle a shareholder class-action lawsuit originally filed in 2020 that accused the company of misleading investors about the legal risks to its contractor-based business model ahead of its 2019 IPO.
* Despite years of legal battles over AB5 in California and federal labor rules, Lyft and Uber drivers remain classified as independent contractors in 2026, largely thanks to Proposition 22, which was upheld by the California Supreme Court in 2024.
* The settlement involves no admission of wrongdoing and changes nothing about driver classification, but it clears a major legal overhang for Lyft while delivering payouts to investors who bought shares between March 2019 and early 2020. A Costly Hangover From the IPO Era
Lyft is finally closing the book on one of its longest-running legal headaches. The ride-hailing company confirmed this week that it will pay $272.5 million to settle a federal securities lawsuit dating back to 2020, centered on how it classified its drivers and what it told Wall Street about that risk.
The deal, which still requires preliminary and final approval from a federal judge in California, would rank among the largest shareholder settlements ever for a gig-economy company. Lyft said it denies any wrongdoing or misleading statements, but agreed to settle to avoid the cost, distraction, and uncertainty of a trial that was inching closer after more than six years of motions, discovery, and appeals.
For investors, it is the end of a saga that began even before Lyft turned a profit. For drivers, it is a reminder of how little has fundamentally changed about their employment status — and how much money has been spent to keep it that way. The Backstory: AB5, the IPO, and the 2020 Lawsuit
To understand the lawsuit, you have to go back to 2019. Lyft went public on March 28, 2019 at $72 per share, pitching itself as the friendlier alternative to Uber and a future leader in transportation-as-a-service.
At the exact same time, California lawmakers were finalizing Assembly Bill 5, or AB5, a landmark law that made it much harder for companies to classify workers as independent contractors. Under its ABC test, a worker is presumed to be an employee unless the company can prove otherwise.
Shareholders who sued in early 2020 — including pension funds that bought in around the IPO — alleged Lyft downplayed that threat. Their core claim: Lyft’s IPO prospectus and subsequent filings described AB5 and driver reclassification as merely possible risks, when internal assessments already showed reclassification would devastate its margins and force a fundamental business-model overhaul.
Lyft shares tumbled through late 2019 and early 2020 as AB5 took effect on January 1, 2020, California’s attorney general sued Uber and Lyft to enforce it, and then the COVID-19 pandemic crushed ride demand. Investors claimed tens of billions in market value were wiped out in part because the classification risk was not fully disclosed.
Lyft has consistently fought that narrative, arguing its disclosures ran to hundreds of pages, explicitly flagged AB5, litigation by the state, and the possibility of higher costs, and that no fraud occurred. A district court initially dismissed parts of the case, but the claims were revived in amended complaints and survived key motions to dismiss in 2022 and 2023, forcing Lyft into prolonged discovery. Why Gig Drivers Are Still Contractors in 2026
Ironically, the doomsday scenario at the heart of the 2020 lawsuit — mass reclassification of California drivers as employees — never happened.
Here’s why drivers are still classified as independent contractors today:
First and most importantly,[...]
AndroGuider | One Stop For The Techy You!
Audible Launches AI Features to Explore Book Worlds and Chat With Characters
https://ai4chat-files.s3.amazonaws.com/images/image_1790879104222.jpg TL;DR
* Audible is rolling out a new suite of immersive, AI-powered tools that let listeners track characters, visualize settings, and explore story lore without leaving the app.
* The standout feature is a generative AI chat experience that lets fans converse directly with characters from select books, with responses grounded in the story and author-approved guardrails.
* The features are launching in the Audible app on iOS and Android in the U.S. starting this week, beginning with popular fantasy, sci-fi, and romance titles before a wider expansion. A New Way to Listen Deeper
Audible wants listening to feel less like pressing play and more like stepping inside the book. The Amazon-owned audiobook giant announced a major update to its mobile app that adds immersive companion tools designed to help listeners keep track of complex plots, remember who’s who, and explore the worlds authors build.
Instead of pausing to Google a family tree or rewind 30 minutes to remember a side character, listeners will now get interactive layers that live alongside the narration. The company says the goal is to solve one of audio’s oldest problems: getting lost in sprawling series with dozens of characters, invented places, and dense mythology.
The rollout begins this week for U.S. listeners on iOS and Android, focused initially on a curated set of bestsellers and Audible Originals. Audible says thousands more titles will be added in the coming months, with publisher and author opt-in. Never Lose Track of Characters Again
One of the core additions is a dynamic character tracker.
As you listen, Audible will surface a living guide to the people in the story — including names, relationships, key motivations, and spoiler-aware recaps that update based on how far you’ve listened. If you’re on chapter 22, you won’t see what happens in chapter 30.
For ensemble casts and long-running series, listeners can tap a character’s profile to see when they last appeared, how they connect to other characters, and a quick “catch me up” summary. Audible says the system is designed to avoid spoilers by syncing tightly with your listening position.
This is especially aimed at fantasy and thriller fans, where keeping track of aliases, houses, crews, and timelines can be half the battle. Explore Places, Objects and Imagery
Alongside characters, Audible is introducing an explorable story world layer.
Listeners can browse key locations, factions, magical systems, technologies, and important objects mentioned in the book, complete with visual imagery and short explainers. Think interactive maps for fictional cities, timelines for historical epics, and glossaries for invented languages and lore.
The feature also pulls out vivid imagery from the narration — letting you linger on descriptions of a castle, spaceship, or small-town diner — and dive into background context without interrupting playback. Audio continues while you explore, and your place is automatically saved.
Audible says the content is generated with a mix of AI summarization and human editorial review, using the book text, publisher materials, and author notes as source material. The Headliner: Chat Directly With Characters
The most buzzworthy part of the launch is a generative AI experience that lets fans interact and converse directly with characters.
For enabled titles, listeners will see an “Ask the Characters” or “Chat” option in the app. Tapping it opens a conversational interface where you can ask questions like “Why did you make that choice?” “What are you afraid of?” or “What do you think of [another character]?” The character responds in voice and text, in personality and knowledge[...]
AndroGuider | One Stop For The Techy You! Legato Launches AI Hearing Glasses to Make Hearing Care Accessible and Stigma-Free
https://ai4chat-files.s3.amazonaws.com/images/image_1790879049123.jpg TL;DR
* Legato has launched AI-powered hearing glasses that combine discreet eyewear with directional smart audio to offer a stigma-free alternative to traditional hearing aids.
* The startup is targeting affordability and comfort with a lightweight, over-the-counter design that costs a fraction of prescription hearing aids and requires no clinic visits.
* The launch signals a broader shift in hearing tech toward mainstream wearables, where AI, personalization, and lifestyle design converge to expand access. Why Traditional Hearing Aids Needed a Rethink
For millions of adults with mild to moderate hearing loss, the biggest barriers to better hearing have never been just about sound quality. They have been about price, comfort, and stigma. Traditional prescription hearing aids can cost thousands of dollars per pair, often require multiple audiologist appointments, and still carry a social stigma that keeps people from seeking help for years. Legato was founded to flip that equation, betting that people are far more likely to wear — and keep wearing — hearing assistance if it looks and feels like something they already love: a great pair of glasses. Meet Legato's AI Hearing Glasses
Legato's new device looks at first glance like premium smart eyewear, with classic frames, clear lenses that can be fitted with prescription lenses, and no bulky earpieces. The hearing technology is built invisibly into the temples of the frames. Using beamforming microphones, open-ear speakers, and on-device AI processing, the glasses selectively amplify voices in front of the wearer while reducing background noise from restaurants, streets, and crowded offices. The result is clearer conversations without the plugged-up ear feeling or feedback whistle common to in-ear aids. Smart Audio Powered by AI
At the heart of the launch is Legato's adaptive sound platform. The system uses AI to distinguish speech from noise in real time, automatically adjusting to different environments, from quiet one-on-one chats to busy group dinners. Users can fine-tune the experience through a companion smartphone app, with personalized hearing profiles, directional focus controls, tap-to-boost conversation mode, and Bluetooth streaming for calls, music, and video. The company says the glasses learn user preferences over time, making everyday listening more natural the longer they are worn. Designed for Comfort and All-Day Wear
Comfort was a central design goal. Weighing about the same as standard eyewear, the Legato glasses avoid the ear canal irritation, pressure, and battery anxiety that cause many people to abandon traditional aids. They feature open-ear audio so wearers stay aware of their surroundings, all-day battery life with a portable charging case, touch and voice controls on the frame, and water and sweat resistance for active lifestyles. For people who wear glasses already, Legato says there is essentially no learning curve — just put them on and hear better. An Accessibility Mission Built on Affordability
Perhaps the most disruptive part of Legato's pitch is price. While traditional hearing aids in the U.S. can run $4,000 to $7,000 per pair and are rarely covered by insurance, Legato is positioning its hearing glasses as an affordable, over-the-counter alternative at a fraction of that cost. No prescription, no referral, and no clinic fitting are required. Customers can order online, customize settings at home via the app, and add prescription lenses through partner opticians if needed. The startup frames this as a public health mission: making early intervention accessible before mild loss becomes social isolation. What This Launch Means for the Futu[...]
AndroGuider | One Stop For The Techy You!
Photon Raises $4.5M to Replace Mobile Apps With AI Messaging Agents
https://ai4chat-files.s3.amazonaws.com/images/image_1790879003490.jpg TL;DR
* Photon has raised $4.5M in seed funding to build a platform that lets developers deploy AI agents natively inside iMessage, SMS/RCS, and email with no app download required.
* The raise was unveiled with a mock funeral for mobile apps in San Francisco, underscoring the startup's bet that conversational agents will replace most consumer app interactions.
* Investors say the approach solves distribution and retention by meeting users where they already chat, positioning messaging as the next app store for AI. A Funeral for Mobile Apps
Photon didn't just send a press release to announce its $4.5M seed round. It staged a funeral.
This week in San Francisco, the startup rolled a black coffin down the sidewalk, complete with flower arrangements, a brass band, and a tombstone reading RIP Mobile Apps 2008-2026. Inside the coffin: a giant iPhone loaded with crowded home screens of forgotten apps.
It was theater, but with a point. Photon CEO and co-founders told onlookers that the average American spends hours a day in messaging inboxes but downloads less than one new app per month. Their argument is that the app store model is exhausted, and AI agents delivered through conversation are what comes next.
The stunt worked. Crowds stopped to film, and the message carried straight into Photon's pitch: why make consumers download, sign up, and learn another interface when they can just text? What Photon Actually Builds
Photon is a developer platform for building and deploying AI agents that live entirely inside native messaging channels — iMessage, SMS/RCS, and email.
Instead of building an iOS app, an Android app, and a web app, a developer can use Photon's APIs, agent builder, and integrations to launch a textable agent in minutes. A customer can book a flight, reorder groceries, reschedule a dentist appointment, get customer support, or shop a drop just by sending a message.
Under the hood, Photon handles the messy parts that have kept brands out of messaging. That includes carrier compliance for SMS, RCS provisioning, Apple Business Chat and iMessage registration, identity verification via phone number and email, payments via Apple Pay and link-based checkout, and memory across channels so a conversation started over SMS can continue over email.
The company also provides agent guardrails, human handoff tools, analytics, and one-click connections to Shopify, Stripe, Zendesk, Salesforce, and calendars. Pricing is usage-based, tied to active conversations and messages resolved, a model Photon says aligns better for businesses than per-seat SaaS. Who Is Betting $4.5M on Chat Over Download
The $4.5M seed round, which brings Photon's total funding to just over $5M including pre-seed, was led by early-stage enterprise and consumer investors with participation from angels in messaging, mobile, and commerce.
In interviews, investors backing the round said they were drawn to two things: distribution and behavior. Messaging has open rates above 90%, dwarfing email and push notifications, and requires zero download friction. If agents can reliably complete tasks, they argue, consumers will overwhelmingly choose to chat instead of tap through five screens.
Photon says it already has dozens of design partners and early customers spanning e-commerce, travel, local services, fintech, and restaurants. Early use cases include concierge shopping agents, AI support agents that resolve tickets over text, and marketing agents that drive reorders via personalized RCS messages. Why Now: RCS, AI Agents, and App Fatigue
Photon's timing is deliberate. Three shifts have converged in the past year.
First, RCS has finally gone mainstream in the U.S. afte[...]
AndroGuider | One Stop For The Techy You!
Brian Chesky: Why AI Agents Need Their Own Operating System
https://ai4chat-files.s3.amazonaws.com/images/image_1790878946321.jpg TL;DR
* Airbnb CEO Brian Chesky says Airbnb is being rebuilt to be agent-friendly, so AI assistants can search, plan, and book stays and experiences on behalf of users.
* Chesky argues consumer AI is still in its dial-up era, with ChatGPT as the only true breakout hit and a massive gap in personalized, trusted, real-world apps.
* His big bet is a new AI-native operating system for autonomous agents, with built-in identity, payments, and trust to let bots transact safely without human clicks. Airbnb's Agent Pivot
In this exclusive interview, Airbnb CEO Brian Chesky laid out the most aggressive tech shift in the company's 18-year history. Airbnb, long defined by its human-centric search box and photo-heavy listings, is being re-architected for a world where the customer isn't always human.
Chesky's vision is simple: whether you type, talk, or delegate to an AI agent entirely, you should be able to Airbnb anywhere. That means opening up Airbnb's inventory of more than 8 million homes and its relaunched Experiences and Services to ChatGPT, Claude, Gemini, and a coming wave of personal assistants.
Rather than fighting agents, Chesky wants to embrace them. Travel planning is messy, multi-step, and deeply personal — the perfect test case for autonomous AI. In his telling, if an agent can handle flights, a week-long itinerary in Tokyo, a dog-friendly cabin, and a last-minute cancellation, it can handle anything. Inside Airbnb's Agent-Friendly Playbook
Chesky detailed a three-part plan already underway inside Airbnb in 2026. First is a complete overhaul of the backend into structured, machine-readable data. Second is a new set of APIs and Model Context Protocol integrations that let third-party agents query live availability, pricing, house rules, and reviews without scraping.
Third, and most visible to users, is Airbnb's own AI assistant built directly into the app. Launched first as AI customer service and now expanding into trip planning, the assistant remembers preferences, travelers, budgets, and past trips. Ask for a quiet beachfront place with fast Wi-Fi for two weeks in November, and it narrows thousands of options to a handful — then books, manages changes, and contacts hosts for you.
Chesky was blunt: Airbnb missed mobile search and had to catch up. It will not miss agentic search. The State of Consumer AI According to Chesky
Beyond Airbnb, Chesky offered a sobering take on consumer AI. Despite billions in investment and hype around foundation models, he believes we have seen only one true consumer breakthrough so far: ChatGPT.
Everything else, he argued, feels like a demo. The models got incredibly smart incredibly fast, but the interface layer — the apps people actually love and use daily — hasn't caught up. It's 1995 all over again, he said, with powerful infrastructure but no Google, no Facebook, no Airbnb yet for the AI era.
The problem is personalization and trust. Current assistants don't really know you, don't remember you across apps, and can't reliably take real-world actions involving money, identity, and safety. Until that changes, mainstream users will remain curious but cautious. Why Agents Need Their Own Operating System
That gap is why Chesky believes the world needs a fundamentally new, AI-native operating system — not for humans, but for agents.
iOS and Android were designed for fingers and eyes: app icons, taps, swipes. Agents don't need screens. They need a system built for delegation, where thousands of bots can discover services, verify who they are dealing with, negotiate, pay, and resolve disputes autonomously.
Chesky outlined what that OS would require: persistent agent identity, so you know which bot boo[...]
AndroGuider | One Stop For The Techy You!
Elon Musk Tesla Abundance Vision Still Vague One Year Later
https://ai4chat-files.s3.amazonaws.com/images/image_1790878894685.jpg TL;DR
* One year after Tesla published Master Plan Part IV centered on "sustainable abundance," Elon Musk still hasn't defined what abundance means in products, timelines, or financial terms.
* Tesla's robotaxi pilot in Austin and Optimus humanoid demos have progressed slowly, leaving the abundance narrative to carry valuation without concrete milestones.
* For investors, abundance remains a vision story rather than a plan, raising stakes for the November shareholder vote on Musk's $1 trillion pay package and 2026 delivery targets. What Musk Promised: Sustainable Abundance For All
In September 2025, Tesla quietly posted Master Plan Part IV on X. Unlike previous master plans that laid out specific vehicles — the Model 3 and Model Y ramp, the $25,000 car, the 20 million vehicle target — this one was philosophical.
Co-authored in Musk's voice, it promised to deliver "sustainable abundance" through AI, robotics, and energy. The document argued that if Tesla could scale autonomous driving, humanoid robots, and solar-plus-storage to near-infinite levels, scarcity in labor, transport, and electricity would effectively disappear.
Musk amplified the message on earnings calls and at the 2025 annual meeting, calling Tesla not a car company but an engine for a post-scarcity civilization. He tied abundance directly to two bets: a global robotaxi network powered by Full Self-Driving, and mass production of the Optimus humanoid robot. No unit economics, no factory plans, no regulatory roadmap were included. Why No Clear Definition Has Emerged Since
Twelve months later, that definition still hasn't arrived. Musk has repeated the word abundance dozens of times — on X, in interviews, during Tesla's Q1 and Q2 2026 earnings calls — but has not explained how it will be measured.
There is still no abundance white paper, no investor day deep-dive, and no updated Master Plan supplement. Tesla's 2025 10-K and 2026 quarterly filings mention Optimus and Robotaxi as growth opportunities but offer no production guidance, pricing model, or capex breakdown tied to abundance.
Analysts who cover Tesla say the vagueness appears intentional. By keeping abundance broad, Musk can fold every project — Dojo, xAI's Grok integration, Megapack, even the delayed Roadster — under one narrative without committing to deliverables that Wall Street could grade him against. When pressed by analysts in July 2026 about when abundance would show up in revenue, Musk replied that people would have to experience it to understand it. Robotaxi: A Pilot, Not A Network
The clearest test of abundance was supposed to be robotaxi. In June 2025, Tesla launched a small supervised pilot in Austin with 10 to 20 Model Y vehicles and safety monitors in the passenger seat.
By October 2026, the service has expanded to parts of the Bay Area and Austin with driverless operation in limited geofenced zones, and Musk claims millions of autonomous miles driven. But Tesla still hasn't disclosed take rates, cost per mile, accident data compared to human drivers, or when the purpose-built Cybercab without steering wheel will enter volume production.
Waymo now operates over 2,000 vehicles across multiple U.S. cities with fully driverless paid rides, while Tesla remains in a testing phase that relies on its vision-only Full Self-Driving stack awaiting broader regulatory approval. Without a scalable, unsupervised network, the abundance claim of nearly free transport remains theoretical. Optimus: Demos Without Deliveries
The second pillar, Optimus, is even less concrete. Musk originally suggested Optimus could begin limited production in 2025 and reach millions of units per year by 2028-2030, potentially accounting f[...]
AndroGuider | One Stop For The Techy You!
California Robotaxi Law Fines Operators for Blocking First Responders
https://ai4chat-files.s3.amazonaws.com/images/image_1790921768745.jpg TL;DR
* California's new first-responder safety law lets the DMV fine robotaxi companies up to tens of thousands of dollars per incident when an autonomous vehicle blocks, delays, or interferes with police, fire, or EMS operations.
* Operators like Waymo, Cruise, and Zoox must now provide 24/7 live emergency contacts, equip cars to pull over automatically for sirens and road closures, and file detailed incident reports within hours.
* The law takes effect January 1 and gives the state power to suspend testing and deployment permits for repeat offenders, raising compliance costs as robotaxi fleets expand in San Francisco, Los Angeles, and beyond. What The New Law Actually Requires
California is putting robotaxi operators on notice: if your driverless car gets in the way of a fire truck, ambulance, or police cruiser, you will pay for it.
The legislation, signed in late September by Gov. Gavin Newsom as part of a broader autonomous vehicle safety package, creates the state's first dedicated penalty system for AV interference with emergency response. Under the rules, which will be enforced jointly by the Department of Motor Vehicles and the California Public Utilities Commission, companies can face fines starting at $5,000 for a first documented obstruction and rising to $25,000 or more per incident for repeated or egregious violations.
But it goes far beyond fines. Operators must now maintain a dedicated, around-the-clock emergency communications line that first responders can reach in under 30 seconds. When a call comes in, companies must be able to remotely locate, unlock, immobilize, or move the vehicle in real time.
The law also mandates technical upgrades. Autonomous vehicles operating for hire must be able to detect active emergency scenes, flashing lights, sirens, road flares, and temporary closures, and then pull over, yield, or reroute without human intervention. If the car cannot proceed safely, it must clear the travel lane rather than stop in place - a direct response to high-profile cases of robotaxis freezing in intersections.
Finally, any interference event must be reported to the DMV within 24 hours, with full sensor video, GPS logs, and remote-assistance records preserved for at least a year. Three major violations in 12 months can trigger a temporary suspension of a company's driverless deployment permit in the affected city. Why Lawmakers Pushed It Through Now
The bill sailed through Sacramento with bipartisan support after two years of mounting complaints from city fire departments, police unions, and mayors.
San Francisco Fire Chief Dean Crispen and Los Angeles County officials testified to dozens of incidents since 2023 where Waymo and former Cruise vehicles stopped in the middle of active fire scenes, parked on top of fire hoses, blocked ambulances trying to reach hospitals, or drove through yellow crime-scene tape while officers were investigating.
In one widely cited San Francisco case, a driverless car stalled between a fire engine and a burning building, delaying crews by several minutes. In Los Angeles, first responders reported having to smash a window to move a robotaxi that would not respond to verbal commands.
Lawmakers said existing rules were toothless. Previously, the DMV could investigate AV collisions, but there was no specific fine for non-crash interference, and cities had no authority to ticket a car with no driver. Emergency responders had no reliable way to contact operators in the moment, often waiting on hold with customer support lines while an incident unfolded.
Supporters framed the law as a compromise. It does not ban driverless operations or give cities veto power over expansion - something th[...]
AndroGuider | One Stop For The Techy You!
MyMonthlyCar Wants to Turn Idle Dealership Inventory Into Rental Revenue at TechCrunch Disrupt
https://ai4chat-files.s3.amazonaws.com/images/image_1790900922769.jpg TL;DR
* MyMonthlyCar, founded by Igor Dobrianskyi, lets dealerships list unsold inventory for flexible month-to-month rentals instead of letting cars sit idle on lots.
* The startup was selected for the Startup Battlefield 200 at TechCrunch Disrupt, taking place October 13 to 15 in San Francisco.
* The model aims to boost dealership profitability while giving drivers a subscription-style alternative to leasing and traditional car rental. The $1 Trillion Parking Lot
Every dealership has them — rows of brand-new and lightly used cars baking in the sun, depreciating by the day while insurance, flooring costs, and lot fees pile up. In a high-interest-rate market where buyers are hesitating and used-car prices remain volatile, idle inventory has become one of the biggest drags on dealership profitability.
MyMonthlyCar wants to turn that liability into a revenue stream. Instead of waiting months for the right buyer, dealers can put those same VINs to work as flexible monthly rentals, generating cash flow today while keeping the retail sale option open for tomorrow. Meet the Founder Behind the Idea
MyMonthlyCar was founded by Igor Dobrianskyi, a mobility entrepreneur who saw firsthand how rigid traditional car access has become. Leases lock consumers in for 24 to 39 months. Daily rentals are too expensive for long-term needs. And dealerships, despite sitting on massive fleets, had no simple way to participate in the subscription economy.
Dobrianskyi's pitch is simple: dealerships already have the cars, the service bays, the insurance relationships, and the local customer trust. What they lack is the software layer to offer those cars the way consumers now want to consume everything else — monthly, flexible, and fully digital. From Lot to Listing in Minutes
The MyMonthlyCar platform is built as a turnkey SaaS and marketplace combo for dealers. A dealership onboards its unsold or slow-moving inventory, sets a monthly price that includes maintenance, roadside assistance, and flexible mileage packages, and publishes the cars to local renters browsing for month-to-month options.
For consumers, there is no long-term lease, no large down payment, and no haggling over financing. They pick a car, verify their license and insurance eligibility online, pay one monthly fee, and can swap, return, or eventually buy the vehicle. For dealers, the platform handles bookings, payments, background checks, rental agreements, and fleet availability in one dashboard.
Crucially, a car rented through MyMonthlyCar is not removed from retail circulation. If a buyer wants to purchase it, the dealer can pull it from the rental pool and close a sale — having already earned rental income on it in the meantime. Why TechCrunch Disrupt Matters
That vision earned MyMonthlyCar a coveted spot in the Startup Battlefield 200 at TechCrunch Disrupt, happening October 13 to 15 in San Francisco. The Battlefield is TechCrunch's flagship showcase for early-stage startups deemed to have high growth and disruptive potential, with past alumni including Dropbox, Mint, and Cloudflare.
For MyMonthlyCar, the timing is strategic. Disrupt puts the company in front of thousands of investors, automotive executives, and press just as dealerships head into year-end with pressure to clear 2026 model-year inventory. A strong showing in San Francisco could accelerate dealer partnerships across California, Texas, and Florida — states where the startup is reportedly focusing its early rollout — as well as unlock seed or Series A funding to scale operations. A Lifeline for Dealership Profitability
The economics are what make dealers listen. A $35,000 SUV sitting unsold f[...]
AndroGuider | One Stop For The Techy You!
OpenAI Fires 3 Safety Researchers Over Sensitive Data Mishandling, WSJ Reports
https://ai4chat-files.s3.amazonaws.com/images/image_1790900869165.jpg TL;DR
* OpenAI has dismissed three safety researchers after an internal investigation concluded they mishandled sensitive company information, first reported by The Wall Street Journal.
* The investigation reportedly centered on the improper sharing and handling of confidential safety-related materials, not on leaks to the press or rival labs.
* The firings come at a critical moment for OpenAI as it races to commercialize advanced models while trying to maintain credibility on AI safety and internal governance. WHAT HAPPENED
OpenAI has parted ways with three researchers on its safety teams following an internal review that found they mishandled sensitive company information.
The dismissals were first reported by The Wall Street Journal, which cited people familiar with the matter. According to that reporting, the researchers were terminated after OpenAI's internal investigation determined company policies for handling confidential information had been violated.
OpenAI has not publicly named the individuals involved and has declined to share detailed specifics about the case, citing employee privacy and security concerns. The company confirmed to the Journal that the three employees are no longer with the firm.
The move was swift and decisive, underscoring how seriously OpenAI leadership is treating internal information controls as the company scales and faces intensifying scrutiny from regulators, partners, and the public. WHAT WAS ALLEGEDLY MISHANDLED
At the heart of the case is not a public leak, but internal mishandling of sensitive materials.
According to WSJ reporting, the investigation found the researchers improperly handled confidential company information related to their safety work. Sources described the issue as a breach of internal data-handling protocols, involving the sharing or storage of sensitive documents outside approved channels.
There is no indication so far that the information was sold, provided to a competitor such as Anthropic or Google DeepMind, or published publicly. People familiar with the matter framed it as a violation of trust and process rather than corporate espionage.
Still, inside a frontier AI lab, even internal mishandling can be high-stakes. Safety research often touches on unreleased model capabilities, red-teaming results, alignment techniques, vulnerability assessments, and risk evaluations — material that OpenAI considers among its most sensitive assets. WHY OPENAI ACTED NOW
The firings reflect a broader hardening of OpenAI's internal security posture.
Over the past two years, OpenAI has dramatically tightened controls on information access, compartmentalized safety and capabilities research, and expanded insider-threat monitoring. The company, now valued in the hundreds of billions of dollars and working closely with Microsoft, government agencies, and enterprise customers, faces enormous pressure to prove it can safeguard intellectual property and pre-deployment safety data.
Leadership has also been eager to project discipline after a turbulent period marked by high-profile departures from its safety and alignment teams, the dissolution of its original superalignment team, and public criticism from former employees who claimed safety was taking a backseat to product launches.
By acting quickly to terminate employees found to have violated policy, OpenAI is sending a clear message internally and externally: safety credentials mean little without operational security to back them up. FALLOUT INSIDE THE SAFETY COMMUNITY
Inside OpenAI and the wider AI research community, news of the dismissals has landed uneasily.
Safety researchers operate in a uniquely difficult[...]
AndroGuider | One Stop For The Techy You!
Fervo Energy Completes World's First Enhanced Geothermal Plant in Record 23 Months
https://ai4chat-files.s3.amazonaws.com/images/image_1790900812073.jpg TL;DR
* Fervo Energy has completed construction of Cape Station Phase 1 in Beaver County, Utah in just 23 months, creating what the company calls the world's first large-scale enhanced geothermal power plant.
* The project uses horizontal drilling, hydraulic stimulation, and real-time fiber-optic monitoring borrowed from the shale industry to unlock heat in hot, dry rock that traditional geothermal couldn't reach.
* With 100 MW ready for grid connection in the first phase and 400 MW more already under development, Fervo says standardized design and faster drilling will make future phases even quicker to build and connect. A Two-Year Build That Changes The Timeline For Clean Firm Power
Fervo Energy has finished building its first full-scale enhanced geothermal power plant in less than two years. The project, Cape Station Phase 1 in Beaver County, Utah, went from groundbreaking to completion in 23 months, a pace almost unheard of for a first-of-a-kind geothermal development.
For an industry where conventional geothermal projects often take 5 to 10 years of exploration, permitting, and construction, the 23-month timeline is the story. Fervo broke ground in late 2023 and early 2024, drilled dozens of deep wells through 2024 and 2025, and completed the power plant and surface facilities by fall 2026. The first 100 megawatts are now in commissioning and preparing to connect to the grid.
Once online, Cape Station will deliver 24/7 carbon-free electricity, enough in its first phase to power roughly 70,000 homes, with the full 500-megawatt station designed to power up to 350,000 homes. How Fervo Pulled Off The 23-Month Sprint
Fervo didn't get faster by cutting corners. It got faster by treating geothermal like shale oil.
Instead of hunting for rare natural steam reservoirs like traditional geothermal, Fervo creates its own. The company drills deep vertical wells 8,000 to 10,000 feet down into hot granite, then turns horizontally for 3,000 to 5,000 feet. Water is pumped down at high pressure to create a network of tiny fractures in the rock, and a second horizontal well recovers the superheated water to drive turbines at the surface.
Three innovations made the speed possible: First, factory-style drilling. Fervo used the same horizontal drilling rigs, polycrystalline diamond bits, and supply chains perfected in the Permian Basin. At Cape, drilling times fell dramatically from well to well, with later wells drilled up to 70% faster than early ones. Second, real-time subsurface control. Fiber-optic cables cemented into the wells provide second-by-second temperature, flow, and fracture data. That let engineers steer fractures precisely and prove reservoir performance without years of test wells. Third, parallel construction. While drilling continued on multi-well pads, Fervo built the power plant, transmission spur, water systems, and substations simultaneously using modular organic Rankine cycle equipment. Early power purchase agreements and streamlined permitting on private land and pre-approved federal zones also removed typical multi-year delays. Why Enhanced Geothermal Is Different
Traditional geothermal only works in a handful of volcanic hotspots like Iceland, California's Geysers, or Nevada. Enhanced geothermal systems, or EGS, work anywhere rock is hot enough at depth — which is almost everywhere if you drill deep enough.
That makes EGS a firm clean energy source: it runs day and night, winter and summer, regardless of wind or sun. Unlike nuclear, it is modular and can be sited in 100-megawatt increments. Unlike natural gas with carbon capture, it has no fuel costs and near-zero emissions once built.
At Cape [...]
AndroGuider | One Stop For The Techy You!
Google Says SpaceX Starship Needs 1800 Launches to Make Space Data Centers a Reality
https://ai4chat-files.s3.amazonaws.com/images/image_1790900743179.jpg TL;DR
* Google has launched its first TPU-equipped prototype satellites into orbit under Project Suncatcher to test whether AI workloads can run in space using direct solar power.
* Google estimates building gigawatt-scale orbital data centers would require around 1,800 SpaceX Starship launches, thanks to Starship's massive payload capacity and planned launch cadence.
* Space-based computing could bypass Earth's power grid bottlenecks with constant, 8x more productive solar energy, but radiation, networking, cost, and debris challenges remain unsolved. Google's First Step Toward Data Centers in Orbit
Google has just taken its most concrete step yet toward moving AI computing off the planet.
The company confirmed it has launched prototype satellites carrying its custom Trillium TPUs into low Earth orbit, kicking off a project internally called Suncatcher. The mission, developed in partnership with satellite imaging company Planet, is designed to answer a deceptively simple question: can modern AI chips survive and operate reliably in space?
The payloads are not full data centers. They are testbeds packed with sensors to measure radiation impact, thermal performance, and compute reliability in the harsh environment of orbit. For now, Google is running ML inference workloads and stress tests, with results expected to shape a much larger constellation planned for late 2027.
If successful, it would mark the first time Google-grade AI accelerators have operated as part of a scalable orbital compute architecture, rather than as one-off space experiments. Why Space? Earth's Power Grid Is Maxed Out
The motivation has little to do with science fiction and everything to do with electricity.
AI demand is exploding. Google says its data center power needs are growing far faster than new grid capacity and clean energy can be built in the U.S. and Europe. Permitting delays, transformer shortages, land constraints, and local opposition are slowing terrestrial expansion.
Space offers a workaround. Above the atmosphere, solar panels receive uninterrupted sunlight with no clouds, no night in sun-synchronous orbits, and about eight times the energy productivity per panel compared to the ground. Instead of competing for grid power, orbital data centers could harvest solar power directly and beam only the results back to Earth.
In Google's vision, future constellations would fly in sun-synchronous orbits, staying in near-constant daylight while machine learning training and inference run on solar power that would otherwise be untapped. Why Google Says It Will Take 1,800 Starship Launches
Even Google admits the math only works with SpaceX's Starship.
In a technical paper accompanying the Suncatcher announcement, Google researchers calculated that a meaningful orbital AI system would need to scale to gigawatts of compute capacity to compete with today's terrestrial hyperscale campuses. That means launching tens of thousands of solar arrays, TPUs, batteries, optical terminals, and structural buses.
Only Starship, with its projected 100-150 ton reusable payload capacity and eventual goal of multiple launches per day, makes that mass-to-orbit affordable. Google estimates roughly 1,800 Starship flights would be required to deploy its first-generation fleet — a number that sounds staggering until you consider SpaceX's target launch rate.
Google explicitly tied the project's viability to Starship achieving rapid reusability and launch costs falling below $200 per kilogram. At Falcon 9 prices, the concept would be economically impossible. At Starship scale, the company argues, launching a 10-megawatt prototype and then scaling to gigawatts be[...]
AndroGuider | One Stop For The Techy You!
ChatGPT Virtual Try-On Launches: OpenAI's New AI Shopping Features Explained
https://ai4chat-files.s3.amazonaws.com/images/image_1790900683161.jpg TL;DR
* OpenAI has launched Virtual Try-On in ChatGPT, letting U.S. shoppers visualize clothing and accessories on their own photos before buying.
* A new Favorites library and personalized For You feed let users save products, get price-drop alerts, and receive tailored recommendations.
* The update signals OpenAI's bigger push into agentic commerce, turning ChatGPT from a chatbot into a full AI shopping assistant powered by merchant integrations. How It Works: Your Photo Becomes the Fitting Room
OpenAI is turning ChatGPT into a virtual fitting room. The new try-on feature, rolling out this week in the U.S., lets shoppers see how an outfit might actually look on them without ever leaving the chat window.
When you ask ChatGPT for shopping ideas - like fall jackets under $200, wedding guest dresses, or new sunglasses - eligible clothing and accessory results now include a "Try it on" button. Tap it, upload a full-body photo or snap one with your camera, and ChatGPT generates a realistic preview of you wearing the item.
OpenAI says the system uses its advanced image generation to preserve body shape, pose, and lighting while draping the garment naturally. You can try on multiple sizes and styles, mix and match tops and bottoms, and ask follow-ups like "show me this in a more casual style" or "will this work for a rainy commute?" For accessories like handbags, sunglasses, and hats, the preview overlays the product on your photo to show scale and fit.
The company is positioning it as a solution to the biggest pain point of online shopping: returns due to poor fit or unexpected looks. Early demos show surprisingly detailed fabric folds, shadows, and proportions, though OpenAI cautions it's a visualization aid, not a guaranteed fit predictor. How to Try It On: A Step-by-Step Guide
Getting started takes seconds, no separate app required:
1. Ask ChatGPT to shop: Type something like "find linen shirts for summer under $75" or "show me trendy sneakers."
2. Look for the hanger icon: On supported products, you'll see a Try it on prompt on the product card.
3. Add your photo: Upload an existing clear, front-facing photo or take a new one. For best results, use good lighting, a neutral background, and fitted clothing.
4. Preview and refine: ChatGPT will generate your try-on in about 20-30 seconds. You can then ask it to swap colors, compare two jackets side-by-side, or complete the look with matching pants.
5. Buy when ready: If you like what you see, tap through to the merchant to check out. U.S. users can also use Instant Checkout for select sellers directly in ChatGPT.
OpenAI says your photos are handled with privacy in mind and are not used to automatically identify you, and you can delete try-on images at any time from settings. The feature is launching first for adult users in the U.S. on iOS, Android, and web, with global expansion planned. Meet Favorites and For You: ChatGPT Gets a Memory for Shopping
Try-on isn't the only upgrade. OpenAI also introduced two new shopping organization tools designed to make ChatGPT feel more like a personal stylist.
Favorites is a new saved-products library built into ChatGPT's sidebar. Anytime you see something you like, tap the bookmark icon to save it with its price, merchant, image, and AI-generated summary. Your Favorites sync across devices, so you can save a jacket on your phone and try it on later from your laptop.
Even more useful: ChatGPT will now send proactive price-drop alerts for items in your Favorites, letting you know when that $180 coat drops to $130.
Alongside that is a new For You feed on the ChatGPT home screen. Similar to TikTok or Instagram shopping tabs, it surfaces c[...]
AndroGuider | One Stop For The Techy You!
Grok Told Trump to Capture Maduro? Inside AI Role in Venezuela Operation
https://ai4chat-files.s3.amazonaws.com/images/image_1790900626965.jpg TL;DR
* President Trump reportedly consulted Elon Musk's Grok AI for strategic scenarios before authorizing the U.S. operation to capture Venezuelan President Nicolas Maduro in early January.
* Grok allegedly outlined capture vs. strike options, risks of regional fallout, and legal justifications, which Trump aides say helped shape the final go-ahead.
* The revelation has ignited debate over AI influence in the White House, data security, and whether chatbots should play any role in life-and-death military decisions. What Reportedly Happened Before the Maduro Raid
According to multiple insider accounts that surfaced in the weeks after the stunning U.S. operation in Caracas, President Donald Trump turned to an unlikely advisor in the final hours before giving the order: Grok, the AI chatbot built by Elon Musk's xAI.
The operation, which resulted in the capture of Venezuelan President Nicolas Maduro and his wife in the early hours of January 3rd, was already months in planning by the Pentagon and intelligence agencies. But as Trump weighed whether to greenlight the high-risk nighttime raid, he reportedly asked aides to query Grok for a fast, plain-English breakdown of outcomes.
Sources familiar with the discussions say Trump was shown Grok responses on his phone at Mar-a-Lago, asking follow-up prompts about Venezuelan air defenses, likely Russian and Chinese reactions, and what would happen if Maduro was killed instead of captured. Musk, who was present for some of the deliberations and whose company has a growing federal footprint, is said to have encouraged the use of Grok as a second opinion to traditional briefings.
The White House has not denied the account, saying only that the President uses many sources of information when making national security decisions. What Grok Allegedly Recommended
Leaked summaries of the exchange suggest Grok did not give a simple yes or no. Instead, it laid out three scenarios: a surgical capture mission, a standoff airstrike campaign, and continued covert pressure.
Grok reportedly flagged a capture mission as high-reward but high-risk, warning of potential U.S. casualties, international legal challenges, and chaos if Venezuela's military fractured afterward. It also warned that airstrikes alone would likely fail to remove Maduro and could rally domestic support around him.
Crucially, Grok is said to have emphasized speed, surprise, and information dominance — recommending a nighttime helicopter-led raid paired with cyber blackouts and immediate public messaging to prevent a power vacuum. That closely mirrors how the actual operation unfolded, with U.S. Delta Force operators, air support, and a communications blackout over Caracas.
Former officials stress Grok did not make the decision. Trump did. But the chatbot's concise, confident tone reportedly resonated with the President, who has long expressed frustration with lengthy Pentagon briefing papers. Why Trump Would Ask a Chatbot About War
For tech insiders, the move is less surprising than it sounds. Trump has publicly praised Grok as fast, uncensored, and loyal, in contrast to what he calls woke AI from rivals. He has a close working relationship with Musk, who spent millions supporting his reelection and briefly led the Department of Government Efficiency initiative.
Inside the White House, Grok has quietly become a fixture. Staffers use it to summarize executive orders, draft speeches, model tariff impacts, and game out media reactions. Unlike classified military systems, Grok is instant, conversational, and available on a phone — qualities Trump prefers over formal interagency memos.
Analysts also point to a broader shift: generals bring cavea[...]
AndroGuider | One Stop For The Techy You!
Kevin Mandia's $2.5B Armadin Startup Uses AI Agent Swarms to Reinvent Enterprise Security
https://ai4chat-files.s3.amazonaws.com/images/image_1790900564061.jpg TL;DR
* Former Mandiant CEO Kevin Mandia has launched Armadin, raising $255.5M at a $2.5B valuation to build autonomous AI agent swarms for enterprise defense.
* Armadin deploys teams of offensive and defensive AI agents that continuously attack, detect, and remediate networks in real time without human bottlenecks.
* Backed by top-tier Silicon Valley investors, the massive debut signals a shift from human-led SOCs to always-on, machine-scale cyber defense. The Legend Behind Mandiant Is Back
Kevin Mandia doesn't start companies for fun. He builds them when the defensive model is broken.
The former U.S. Air Force officer turned founder of Mandiant defined modern incident response. After FireEye acquired Mandiant for over $1 billion in 2013, Mandia led the combined company, took it back to the Mandiant name, and then steered its $5.4 billion sale to Google in 2022. He stayed to integrate Mandiant into Google Cloud Security, then quietly stepped away.
Now, just as AI-driven attacks are overwhelming enterprise security teams, Mandia is returning with what he calls his final architecture: Armadin.
Announced this week, Armadin has come out of stealth with a staggering $255.5 million in funding at a $2.5 billion valuation, one of the largest cybersecurity seed-to-Series A debuts on record. What Is Armadin Building? Security As A Swarm
Armadin isn't another AI copilot for analysts or a chatbot layered on top of a SIEM. It's a complete rethink: a swarm of autonomous AI agents that live inside the enterprise and fight as a coordinated unit.
The platform fields dozens to hundreds of specialized agents working in parallel. Red agents continuously probe for weaknesses, mimicking ransomware gangs, nation-state APTs, and insider threats. Blue agents monitor telemetry, hunt for anomalies, and triage alerts. Purple agents in the middle validate exploits, kill false positives, and autonomously deploy fixes, from isolating endpoints to rewriting firewall policies and rolling back identities.
Instead of annual penetration tests and alert queues that take days to clear, Armadin promises continuous, live-fire testing and defense at machine speed. Human analysts move from the front line to commanders' intent, setting rules of engagement while the swarm executes.
Mandia has described it as going from castle walls to an immune system. Inside The $255.5M Mega-Raise
The scale of the raise reflects both Mandia's track record and the urgency around AI security. According to the company, the $255.5 million round was raised across stealth seed and Series A tranches, led by Andreessen Horowitz and Sequoia Capital with participation from Lightspeed Venture Partners, ICONIQ Growth, and strategic angels including former CISOs from Fortune 100 banks and tech giants.
The $2.5 billion valuation puts Armadin instantly into decacorn-track territory, ahead of even Wiz and Abnormal at a similar stage. Sources close to the deal say the round was heavily oversubscribed, with investors betting that Mandia can repeat the Mandiant playbook of trust with CISOs and government leaders.
Armadin says it already has design partners in financial services, healthcare, critical infrastructure, and federal-adjacent sectors, and will use the capital to scale its engineering team in Northern Virginia, Austin, and Silicon Valley, plus build sovereign-ready data centers for regulated customers. Why Investors Are Betting Big On Autonomous Defense
The timing is no accident. In 2026, enterprise SOCs are drowning. AI-generated phishing, deepfake social engineering, polymorphic malware, and automated vulnerability discovery have collapsed attacker dwell time from d[...]
Proposition 22. In November 2020, Uber, Lyft, DoorDash and Instacart spent more than $200 million to pass the California ballot measure, which carved out app-based drivers from AB5. Prop 22 kept drivers as contractors while promising a minimum earnings guarantee, health-care stipends for high-volume drivers, and accident insurance.
Labor groups challenged Prop 22 as unconstitutional, and a lower court initially agreed. But in July 2024, the California Supreme Court upheld Prop 22 in a decisive ruling, cementing contractor status for nearly a million California gig workers. That ruling effectively ended the state’s AB5 enforcement push against ride-hailing.
Second, a patchwork of other state wins. Following California’s lead, Washington state passed a compromise law preserving contractor status with minimum pay standards, while New York, Massachusetts and Minnesota struck minimum-pay deals for drivers without changing classification. No state has successfully forced Uber or Lyft to reclassify all drivers as employees.
Third, federal whiplash. The U.S. Department of Labor under the Biden administration issued a 2024 rule that made it easier to classify gig workers as employees under the Fair Labor Standards Act, raising alarms across the industry. But enforcement was limited, tied up in court challenges, and later rolled back under the Trump administration in 2025-2026 in favor of a more business-friendly standard. As of October 2026, there is no federal law requiring gig driver reclassification.
The result: Lyft, Uber and DoorDash still operate on a contractor model nationwide, supplemented by Prop 22-style benefits in a few states rather than full employee wages, overtime, unemployment insurance, and expense reimbursement. Inside the $272.5 Million Settlement
The proposed $272.5 million payout covers investors who purchased Lyft Class A common stock between its March 2019 IPO and around March 2020, before Lyft disclosed mounting losses and California litigation escalated.
Lawyers for the lead plaintiffs called it an excellent recovery given the difficulty of proving securities fraud, noting Lyft’s stock drop was also driven by the pandemic and competition with Uber — factors Lyft argued broke any causal link.
For context, the amount is substantial but manageable for today’s Lyft. The company, which finally turned consistently profitable on an adjusted basis in 2024 and posted its first full-year GAAP profit in 2025, held more than $1.7 billion in cash and short-term investments as of its last quarterly report. It said it expects to fund the settlement through a mix of cash on hand and insurance proceeds from its D&O policies, and will take a one-time charge in the third quarter of 2026.
The settlement does not allege any current accounting issues and does not require Lyft to change its business practices, disclosures going forward beyond what is already required by securities law, or its treatment of drivers. What It Means for Lyft Going Forward
Wall Street largely treated the news as relief. Legal overhang has weighed on Lyft for years, alongside concerns about autonomous competition from Waymo, Tesla Robotaxi, and Uber’s larger scale.
Analysts say clearing a potential multi-hundred-million-dollar trial verdict removes uncertainty ahead of 2027 and lets management focus on its current playbook: expanding Lyft’s partnerships for autonomous rides, growing its advertising and subscription businesses, holding market share in the U.S. and Canada, and returning capital through buybacks.
The risk is financial, not existential. At $272.5 million, the payout equals roughly a quarter’s worth of gross bookings profit, but it will dent free cash flow and likely pause additional buybacks this year. Lyft also still faces ongoing driver-pay lawsuits, FTC scrutiny over advertising of driver earnings, and state-level minimum-pay fights — none of which are resolved by this deal.
In a statement, Lyft framed the settlement as putting a legacy I[...]
consistent with that point in the story.
Audible emphasizes the chats are grounded in the book’s canon, not open-ended roleplay. The AI is constrained to what the character would know at your current listening progress to prevent spoilers, and it will decline to reveal future plot points. Authors and publishers can set boundaries, review personas, and opt out entirely.
In a demo shared by the company, a listener was able to interrogate a detective about a clue, get pep talks from a romance lead, and ask a fantasy villain to explain their backstory — all with distinct tone and voice styling separate from the main narrator. How It Works Under the Hood
Audible isn’t detailing the exact models involved, but says the experiences run on Amazon’s generative AI infrastructure combined with its own listening data and proprietary story graphs.
Each supported book gets a structured model of its narrative — characters, relationships, arcs, locations, and events — built ahead of time. The live features then query that model based on your timestamp in the audiobook.
The company stresses human oversight: editorial teams and rights holders review AI-generated profiles and chat guardrails for accuracy, tone, and safety. Chat conversations include clear labeling that you’re talking to an AI interpretation, not the author, and include reporting tools and usage limits. Privacy and Control for Authors and Listeners
As with any generative AI launch, control is a central question.
Audible says participation is opt-in for rights holders, with compensation terms covered under existing audio agreements for participating titles. Authors can customize character personalities, approve what can and cannot be discussed, and disable chat for specific characters.
For listeners, chat histories can be viewed and deleted, and the immersive layers can be turned off entirely for a traditional, distraction-free listening experience. Audible also says the features include spoiler protection by default and kid-safe filters for young adult titles. What’s Next for Immersive Listening
This launch is part of Audible’s broader push to differentiate beyond just a catalog of audiobooks. Over the past year, the company has tested AI-generated recaps, personalized recommendations, and faster search — but this is its most ambitious attempt to make the app itself part of storytelling.
Analysts see it as a direct play for fandom engagement, putting Audible in closer competition with platforms like Spotify, TikTok’s BookTok community, and AI companion apps — while giving it a new hook to retain subscribers in fantasy, romantasy, and sci-fi, some of audio’s fastest-growing genres.
Audible says international expansion, more languages, and support for Audible Plus Catalog and library-shared titles are on the roadmap. If fans embrace chatting with their favorite heroes — and villains — expect story worlds to become a standard part of how audiobooks are made, marketed, and experienced.
re of Hearing Tech
Legato's debut arrives as hearing care converges with the broader wearable revolution. Following recent FDA rules expanding over-the-counter hearing options and growing consumer interest in AI earbuds and smart glasses, analysts see hearables as the next major battleground for Apple, Meta, Bose, and startups alike. By merging vision and hearing into a single, socially accepted device, Legato is testing whether design can do what technology alone has not — get tens of millions of untreated adults to finally take action on their hearing. If the company can deliver on sound quality, battery life, and support at scale, its stigma-free approach could redefine what a hearing aid looks and feels like.
r Apple added support on iPhones, upgrading SMS with branded messaging, carousels, high-res images, and suggested replies. That gives developers a far richer canvas without needing an app.
Second, large language models and agent frameworks have matured to the point where an AI can actually take multi-step actions — checking inventory, processing refunds, booking calendars — rather than just answering FAQs.
Third, both consumers and developers are burned out on apps. Acquisition costs have soared, uninstall rates remain brutal, and most startups never crack the App Store top charts. Texting an agent sidesteps that entire funnel.
Photon is positioning itself as the Twilio-meets-Vercel for this era: the infrastructure layer that makes agent-powered messaging as easy as deploying a website. What Comes Next
With the fresh capital, Photon plans to triple its engineering team, expand its RCS and Apple Messaging integrations internationally, and launch a self-serve agent marketplace where businesses can publish discoverable agents with their own phone numbers and email addresses.
The startup will also invest heavily in reliability, compliance, and anti-spam tooling — critical if it wants to avoid the fate of previous waves of SMS startups that overwhelmed users. Founders say every Photon agent must be opt-in, clearly identified as AI, and offer instant human escalation and opt-out.
Competition is heating up, from customer messaging giants adding AI agents to new agent-builder startups focused on web chat and WhatsApp. Photon's differentiator, it claims, is its native focus on the American inbox: iMessage, Android RCS/SMS, and email, rather than social apps.
Whether consumers truly bury mobile apps remains to be seen. But Photon has made its thesis impossible to ignore — coffin and all. If its developers succeed, your next favorite app might not be an app at all. It might just be a contact in your messages.
ked what; native payments and escrow for agent-to-agent transactions; and a reputation and trust layer to prevent fraud, spam listings, and hallucinations. Without it, he warned, we'll get a chaotic web of bots scraping sites and breaking things.
He likened it to the early internet needing HTTP and payments needing Stripe. AI needs its own trust and transaction protocol, and travel — with its high stakes, real money, and real homes — is where it must be proven first. What Comes Next for Airbnb and AI
For Airbnb, the stakes are existential and opportunistic. If OpenAI, Google, or Apple become the default travel concierge, Airbnb risks becoming a dumb commodity backend. By making itself the most agent-accessible platform with the richest structured data and strongest host-guest trust system, Chesky hopes Airbnb becomes the default place agents go.
Expect more partnerships with leading AI labs, deeper integration into voice assistants and chatbots, and an Airbnb app that feels less like a search engine and more like a personal travel agent that knows you.
Chesky's closing message was characteristically ambitious: the next great operating system won't be tapped, it will be told what to do. And he wants Airbnb to be one of its first killer apps.
or the majority of Tesla's long-term value.
A year later, Optimus has appeared in staged videos folding laundry, dancing, and working inside Tesla's Fremont factory, plus a live appearance at company events. Tesla says Gen 3 hardware with Grok-powered voice interaction is in development and that pilot deployment in its own factories will start in 2026.
But there is no price, no external customer, no factory tooling announcement, and no safety certification path for home use. Former Tesla AI staff and robotics experts note that dexterous manipulation, battery life, and cost — Musk has floated a $20,000 to $30,000 price — remain unsolved at scale. Abundance via household robots still lives in keynote slides, not shipping manifests. What It Means For Tesla Investors
For shareholders, the stakes around abundance have only grown. Tesla's auto business faces its second straight year of delivery pressure, with increased competition from BYD and other Chinese EV makers, an aging lineup, and the loss of U.S. EV tax credits weighing on 2026 demand.
Yet the stock continues to trade at a multiple far above traditional automakers, largely on the promise that robotaxi and Optimus will unlock trillions in value. That makes abundance more than branding — it is the justification for the valuation.
The tension comes to a head this fall. Tesla has proposed a new performance-based compensation package for Musk worth up to $1 trillion over a decade, tied in part to achieving massive milestones in autonomous deployment, robotics, and market cap. Proxy advisors and some large funds want specifics: how many robotaxis, how many Optimus units, by when, at what margin. So far, Tesla's board has framed the vote as a bet on Musk delivering abundance, without defining abundance itself. The Road Ahead: Vision Needs A Plan
Musk has never operated like a traditional CEO with five-year guidance tables. His strength has been setting impossible-sounding goals — landing rockets, scaling Model Y to the world's best-selling car — and eventually getting close enough to change industries.
The difference now is that abundance is not a product. Investors can track Model Y deliveries, Megapack installations, and even FSD miles. They cannot track abundance.
Until Tesla translates the vision into factories, unit costs, regulatory approvals, and revenue segments, abundance will remain what critics say it is today: a compelling story to sustain momentum while the hard engineering and business work catches up. For believers, that is enough — Musk's track record earns him time. For skeptics, one year of vagueness is already one year too long.
our expectation of privacy ends the second someone puts on a pair of smart glasses."
The tech industry and free-expression groups applauded the move. The Consumer Technology Association and Meta had lobbied against the bill, arguing that all of its devices already include a visible recording indicator light and that misuse should be punished, not the technology itself. The ACLU of California had also raised concerns that the bill could punish homeless outreach workers, cop-watch activists, and journalists using discreet wearables for accountability reporting.
Meta said in a statement that it supports strong voyeurism laws but that device-specific bans are unworkable when cameras are now in phones, doorbells, watches, pins, and glasses. The Bigger Problem: Smart Glasses Are Going Mainstream
The veto lands as camera-equipped wearables are exploding from niche gadget to mainstream product.
Meta alone has sold millions of Ray-Ban Meta glasses, and its new $799 Ray-Ban Display with a built-in heads-up screen and AI assistant launched this fall to long lines. Google is preparing to re-enter the market with Android XR glasses, while Apple, Amazon, and Snap are all racing toward lighter, more discreet designs with longer battery life and live AI transcription.
Privacy researchers warn the threat is no longer just secret video. Paired with facial recognition workarounds and multimodal AI, a stranger can now identify you, look up your social profiles, and transcribe your conversation in real time — all without pulling out a phone.
Universities, bars, movie theaters, and gyms across California have already started posting no smart glasses signs, but enforcement is nearly impossible because the latest models look exactly like regular prescription glasses.
Critics say the tiny white LED that turns on during recording is easy to cover with tape or miss in daylight, making consent effectively meaningless in crowded public spaces. What This Means For The Future Of Public Surveillance Laws
Newsom's veto does not end the fight — it resets it.
The governor signaled he would support a narrower bill next session focused on behavior, not devices, such as strengthening penalties for recording in locker rooms, restrooms, changing rooms, and for sexual gratification, regardless of whether the camera is in glasses, a phone, or a button.
Lawmakers who authored the measure vowed to reintroduce a revised version in January 2027, potentially with clearer exemptions for journalism, law enforcement bodycams, and assistive technology for the blind and low-vision community.
In the meantime, California — often the state that sets the national standard for privacy with laws like the CCPA — will be left without a smart-glasses-specific statute, just as the 2026 holiday shopping season is expected to put millions more AI wearables on faces.
Legal experts say expect three things next: more lawsuits testing whether existing voyeurism laws apply to wearables, more private bans by businesses, and more pressure on Congress to pass a federal standard for recording indicators, AI facial recognition limits, and public-space privacy.
For now, in California, it's still legal to wear camera glasses in public. What's illegal is what you do with them — a distinction Newsom says must remain clear, but privacy advocates say is already dangerously blurry.
