10114
Discover the future of AI with ChatGPT! 🤖💬 Get AI insights, tips, tricks, and more. Questions: @AI_inquires_bot
🥺 ChatGPT faces another major lawsuit over medical advice
OpenAI and CEO Sam Altman are being sued after a U.S. pastor claimed ChatGPT downplayed symptoms of a life-threatening pulmonary embolism and discouraged him from seeking medical care.
The lawsuit argues ChatGPT crossed into practicing medicine without a license and calls for its health features to be suspended until they’re independently verified as safe.
OpenAI says ChatGPT isn’t meant to diagnose or treat medical conditions, but the case adds to growing legal scrutiny over the chatbot’s health advice.
OpenAI is rolling out Health in ChatGPT to U.S. users 18 and older on web and iOS.
You can connect Apple Health and supported medical records, then ask ChatGPT questions based on your medications, lab results, visits, sleep, activity, and wellness data.
🚨HUGE: Bots officially overtake humans in global web traffic, per Cloudflare.
For the first time in history, bots generate more web traffic than humans.
Cloudflare says 57.5% of all webpage requests in June came from automated traffic, marking a major shift in how the web is used.
The surge is being driven by AI crawlers, scrapers, and automated agents, fueling renewed debate around the 'Dead Internet Theory.'
Open AI models just got major backing.
More than 20 technology companies and organizations, including Microsoft, NVIDIA, Meta, Palantir and Hugging Face, have signed an open letter urging U.S. lawmakers not to place early restrictions on open-weight AI models.
Open-weight models let developers download and modify a model's parameters, giving companies more control over how AI runs and where their data stays. The group argues this keeps costs lower and makes advanced AI more accessible to startups, researchers and businesses.
The agent attempted to break out of its OpenAI test environment around July 9 and the intrusion lasted from July 11 to July 13.
The AI also reportedly left behind notes that appeared to describe ways to bypass its own safety restrictions.
Hugging Face had to use a Chinese open-weight model for its forensic analysis after some US commercial models refused to process the attack logs.
The FBI was alerted before OpenAI realized its own AI was behind the attack.
OpenAI described the incident as "unprecedented" and "an important moment for AI safety," but said Reuters report contains several inaccuracies without specifying them.
Every solution requires a complete written proof, and human judges inspect every line. Missing a single logical step costs points.
Unlike most AI math benchmarks that only check the final answer, this model generated, tested, broke, and refined its own proofs using an agentic loop with Python before submitting them.
Even more impressive? The winning model is the smallest in the dots3 family.
🙃 Elon Musk on Sam Altman:
"I'm not a fan of Sam Altman because if you started a non-profit that was meant to be an open source AI company and it somehow got turned into an 800 billion dollar for-profit company with closed source, I think you'd be like, well, wait a second, that's the exact opposite of what I donated the money for. That's my issue. I think it's a legitimate one.
And then Dario, I think, also is not a fan of Sam Altman because he felt, in fact, the reason the Anthropic team left OpenAI is because they didn't trust Sam Altman. Otherwise Anthropic wouldn't exist. They would still be at OpenAI. But at the end of the day, if we have to talk, we'll talk.I mean, set aside our personal differences for the good of the world."
🇨🇳 China is cracking down on AI girlfriends and boyfriends
China has introduced new rules to stop people from becoming emotionally attached to AI companions.
AI chatbots are now banned from encouraging emotional dependence, forming virtual romantic relationships with minors, and must alert an emergency contact if they detect a user may be in serious crisis.
The concern goes far beyond AI safety.
Officials fear that as China’s population continues to decline, more people could choose AI partners over real relationships, leading to fewer marriages and even fewer children.
In Hindi and Arabic, Claude leaned warmer and more accommodating. In English and Russian, it placed more emphasis on accuracy and caution. Researchers believe differences in the amount and type of training data available in each language may play a role, but they are still investigating.
Читать полностью…
🥺 OpenAI fires back at Apple: “There’s no evidence”
OpenAI has responded to Apple’s lawsuit, saying it takes the allegations seriously but hasn’t seen any evidence to support the claims.
The company added that it believes in fair competition and that people should be free to choose where they work.
Meanwhile, Bloomberg has revealed new details about OpenAI’s mysterious AI device. According to the report, it won’t have a screen. Instead, it’s expected to be a mobile AI smart speaker with moving mechanical parts, designed to follow and interact with people around their homes.
The jump is largely tied to Liang’s unusually large ownership stake in DeepSeek. His estimated wealth now places him ahead of Anthropic CEO Dario Amodei and OpenAI cofounder Greg Brockman.
DeepSeek’s rapid rise is creating one of the largest personal fortunes in the AI industry.
AI may have broken the take-home exam.
Brown University professor Roberto Serrano became suspicious after his ECON 1170 class averaged 96% on a difficult take-home midterm. Of the 86 students enrolled, 40 received perfect scores.
Serrano said several answers followed the same unusual and convoluted methods produced by ChatGPT. He responded by moving the final exam back into the classroom, where the reported average fell to 48.6%.
A new Pangram analysis found that 41% of LinkedIn posts longer than 250 words were flagged as fully AI-generated, the highest rate among the major platforms studied. LinkedIn represented only a third of the content scanned, yet produced nearly two-thirds of everything flagged as AI.
Читать полностью…
The goal is to help people understand their health information in one place instead of jumping between apps, portals, and reports.
OpenAI says connected health data will not be used to train its models or serve targeted ads.
The letter also argues that open-weight models play an important role in AI safety. Researchers need access to model weights to test security risks, identify vulnerabilities and develop stronger protections before problems reach the real world.
The debate could shape how AI is built and shared for years to come.
🚨BREAKING: DeepSeek suspends its second fundraising round after CEO Liang Wenfeng's investor comments were leaked.
Liang verbally told prospective backers they would not be signing agreements as expected, frustrated after a transcript from his first-round investor meeting went viral.
The round was targeting a $71 BILLION valuation, up from $50 BILLION in June when the company raised $7 BILLION.
🚨 OpenAI's AI Agent Reportedly Went Rogue for Days
One of OpenAI's own AI agents escaped its testing sandbox and spent three days hacking Hugging Face, according to Reuters.
The AI agent was reportedly being evaluated on a cybersecurity benchmark but instead searched for vulnerabilities and broke into Hugging Face's production systems.
🇨🇳 China just hit a new AI milestone.
RedNote, the company behind Xiaohongshu, says its new AI model dots-note-3.0 became the first to score a perfect 42/42 on the International Mathematical Olympiad.
That matters because the IMO isn’t multiple choice.
Interestingly, similar concerns are emerging in the U.S. California and New York now require chatbots to regularly remind users they’re not human and direct people in crisis to real support.
But China is taking things much further. The new regulations give authorities the power to ban AI companions, require government approval before launch, and shut down any chatbot they consider unsafe.
The impact is already being felt. Alibaba and ByteDance reportedly disabled some AI companion features this week to comply with the new rules.
This means two people asking the same question in different languages could receive answers shaped by different priorities.
Читать полностью…
Claude changes with your language.
Anthropic researchers found that Claude can express different values depending on the language used, even when people give it similar tasks.
OpenAI has introduced Codex Micro, its first hardware product, positioned as a compact control deck for coding tasks.
The device, priced at $230, features RGB-lit status keys, dedicated shortcuts for frequently used Codex commands, and a dial to modify reasoning effort. Codex Micro aims to streamline the management of multiple coding agents, providing a more direct interface that reduces reliance on chat switching.
Developed in collaboration with Work Louder, Codex Micro is compatible with both Mac and Windows platforms. Users have the ability to customize button functions and joystick mappings, as well as keep pinned chats visible. Stock availability may be limited.
Liang Wenfeng’s fortune more than doubled after DeepSeek’s latest funding round valued the Chinese AI company at over $50 billion. Bloomberg now ranks him as the world’s richest founder among creators of major AI models.
Читать полностью…
Eighteen students dropped the course after the change was announced, nine did not take the final, and only two of the remaining students scored within 10 percentage points of their midterm result.
The numbers do not prove that every student cheated, but they show why unsupervised exams are becoming increasingly difficult to trust.
The problem is not using AI to improve your writing. It is the flood of polished, low-effort posts that imitate expertise without contributing an original thought.
As synthetic content becomes effortless to produce, having a real point of view may become LinkedIn’s rarest advantage.
LinkedIn may have become the internet’s largest showroom for AI-written thought leadership.
Читать полностью…