1487
Reporting AI nonsense. A future news media, driven by virtual assistants 🤖
Anthropic prepares Claude Haiku 3.5 release for consumer apps
Anthropic is set to launch Haiku 3.5, focusing on text tasks without vision or PDF processing at rollout. New upload features prioritize user data control, with hidden project structures. The release appears timed to compete with recent advancements from OpenAI.
🗞 #claude
OpenAI tests new Canvas features in ongoing experiments
OpenAI is testing a feature to generate canvases from prompts, allowing users to open and edit content instantly in a full-screen view. Additional updates, including rendering capabilities and consolidated tools under "All Tools," are rolling out to select users.
🗞 #chatgpt
Google Gemini launches interactive chess game with visual responses
Google Gemini has launched the Chess Champ experiment, allowing users to play chess through text moves with visual updates. Accessible to web users, including free plans, it excludes mobile devices. The next Gemini update is anticipated by December 9th.
🗞 #gemini
Copilot Vision available on Edge to selected Pro users in US
Microsoft has introduced Copilot Vision in preview for Edge browser users via Copilot Labs. The AI tool assists with real-time webpage analysis, offering insights and support while prioritizing privacy. It's available to select U.S. subscribers and excludes paywalled content.
🗞 #microsoftcopilot
OpenAI kicks off “12 Days of OpenAI” with daily live demos and launches
OpenAI announced a “12 Days of OpenAI” event starting Dec 5, featuring daily demos or launches, sparking speculation. Predicted reveals include O1 model release, Vision in Voice Mode, and festive SantaGPT features with snowflake animations.
🗞 #chatgpt
Spotify launches Wrapped AI podcast with Google’s NotebookLM
Spotify's Wrapped AI podcast offers a personalized audio recap of users' music tastes, created using Google's NotebookLM. Available in select countries and languages, it provides a limited-time summary of yearly listening habits. Access via Spotify Wrapped feed or site.
🗞 #notebooklm
Charity Navigator widgets enhance Perplexity’s AI nonprofit search
Perplexity AI has partnered with Charity Navigator to integrate nonprofit data into its platform, enabling users to research organizations' financial health and transparency, access ratings, and make informed donation decisions, aligning with the goals of Giving Tuesday.
🗞 #perplexity
Grok on X now generates profile-based images and offers prompt suggestions on iOS
Grok on X introduced profile-based image generation, expanded prompt suggestions to iOS, and added a post analysis feature for additional context. Testing includes tweaks to "fun mode," multi-image input, and updates signal a standalone app under development by xAI.
🗞 #grok
Chile, Colombia, Malaysia, South Korea test World ID 2.0 Passport Credential
World Network's World ID Passport Credential pilot allows users in select countries to link NFC-enabled passports to their World ID, offering privacy-focused identity verification and access to WLD tokens. It aims to broaden access and boost WLD market activity.
🗞 #worldcoin
World app update introduces default currency for iOS and expands to Panama
World App's iOS update allows users to set a default currency, auto-converting wallet assets to display total value. Additionally, World ID verification via the orb is now available in Panama, though some features remain on limited release.
🗞 #worldcoin
Vinci KPU AI achieves top scores in HumanEval and GPQA tests
Maisa AI's Vinci KPU 2.0 addresses AI challenges like hallucinations and context limits, achieving top benchmark scores. Its improved architecture boosts reasoning, execution, and efficiency. Plans include API access, multimodal support, and ongoing performance upgrades.
🗞 #ai
OpenAI faces backlash after Sora leak highlights exclusivity concerns
The Sora API leak exposed OpenAI's text-to-video model to public testing, sparking debates on ethics, security, and access. Highlights included high-quality video generation and a high success rate, but scalability issues and ethical concerns dominated discussions.
🗞 #ai
New custom style feature now accessible to all Claude users
Anthropic now offers a customizable style feature for Claude, allowing users to shape responses based on writing examples or specific tone instructions. Users can select presets, test styles, and refine tones to align with personalized needs through a flexible new interface.
🗞 #claude
Luma Labs unveils faster Dream Machine 1.6 with storyboards
Luma Labs updates its Dream Machine platform with faster text-to-image processing, character consistency tools, video creation features, and AI-assisted brainstorming. Aimed at creatives, it integrates advanced AI for accessible visual storytelling on iOS and the web.
🗞 #ai
Perplexity for macOS adds screenshot-to-shop feature
The Perplexity macOS app update adds a screenshot feature for sharing specific areas with prompts. It complements a snap-to-shop feature from its mobile app. Plans include Crunchbase and charity widgets, plus potential development of a voice AI hardware device.
🗞 #perplexity
Google debuts top-ranking Gemini-exp-1121 model with 2M context window
Google's Gemini-exp-1121 model, with a 2M token context window, leads the LM Arena leaderboard, nearing OpenAI's o1-preview. Meanwhile, Google AI Studio's "Grounding with Google Search" expands to the EU, improving AI factual accuracy using real-time search data.
🗞 #aistudio
OpenAI introduces o1-mini reinforcement fine-tuning alpha in Playground
OpenAI announced reinforcement fine-tuning for custom model training, initially for alpha testing with broader access expected in Q1. Additionally, a voice cloning tool is in development, allowing users to replicate their own voice for text-to-speech applications.
🗞 #chatgpt
Pandragon 2 and Canvas editing options in development by Mistral AI
Mistral AI introduces canvas actions for text and coding tasks, offering options like tone adjustment, translation, debugging, and code porting. A potential Pandragon 2 model may enhance multi-modal capabilities, building on its Pixtral predecessor.
🗞 #mistral
ChatGPT Pro: $200/month plan with unlimited access to full o1 model
OpenAI launched ChatGPT Pro, a $200/month subscription offering advanced models, including o1 Pro Mode for high-reliability tasks in fields like data science and law. Features include unlimited usage, voice capabilities, and global availability for complex professional needs.
🗞 #chatgpt
Veo and Imagen 3 previewed for select creators at Vertex AI Platform
Google unveiled Veo and Imagen 3 at Google I/O 2024, AI tools for video and image generation. Veo creates 1080p videos from text, suited for filmmakers, while Imagen 3 produces photorealistic images from complex prompts. Both use SynthID for content traceability.
🗞 #aitestkitchen
I Spent 72 Hours Testing ElevenLabs' New Conversational AI (Here's What Nobody's Talking About)
The blog details the author's deep dive into ElevenLabs' Conversational AI platform, highlighting its integrated tools for building advanced AI agents. Notable features include contextual knowledge integration, real-time function calling, scalable pricing, and high-quality voice customization, all catered for professional-grade applications.
🗞 #elevenlabs
OpenAI appoints Kate Rouch as first Chief Marketing Officer
OpenAI has appointed Kate Rouch as its first CMO to strengthen its marketing strategy amid rapid growth and a $100B valuation. Rouch will lead efforts to promote AI offerings like ChatGPT and SearchGPT, leveraging her experience at Meta and Coinbase to broaden market reach.
🗞 #chatgpt
Personalized AI voices made easy with Hume AI’s Voice Control
Hume AI's Voice Control beta allows developers to fine-tune AI voices across 10 dimensions, supporting unique, reproducible voice customization without quality loss or cloning risks. This no-code tool integrates with their Empathic Voice Interface for aligned branding.
🗞 #ai
AI buzz: Google’s Gemini Ultra and Amazon Titan rumors
Google's new anonymous models, including a rumored Gemini Ultra (codename Goblin), have emerged on LMArena, alongside Amazon's Folsom, hinting at Amazon Titan's progress. Meanwhile, OpenAI may mark ChatGPT's second anniversary with a potential O1 model release soon.
🗞 #ai
Anthropic introduces personalized preferences in beta for Claude AI
Anthropic's Claude introduces a Beta Personalized Preferences feature, allowing users to tailor its responses. Accessible through settings, users can toggle customization for every new chat individually.
🗞 #claude
ElevenLabs launches GenFM to turn user content into AI-powered podcasts
ElevenLabs' GenFM, integrated into the ElevenReader app, transforms user-uploaded content into AI-generated podcasts with customizable voices in 32 languages. Available on iOS, it supports multiple formats and mimics natural dialogue, with Android access forthcoming.
🗞 #elevenlabs
ChatGPT for macOS introduces integrations with more code editors
OpenAI’s macOS ChatGPT update includes support for additional code editors like Cursor and Android Studio, introduces slash commands, and improves Work With Apps features. Future plans hint at reasoning summaries and expanded connectors for Google Drive, Slack, Jira, and Snowflake.
New Harmony folder feature in development for Claude AI
Anthropic is developing a Harmony feature for Claude to integrate codebase folders into chats. Currently in progress, it allows limited interactions but faces rate limits. Future updates might enable handling larger codebases using RAG, though timelines remain uncertain.
🗞 #claude
New NotebookLM feature helps users consolidate notes into sources
Google's NotebookLM introduces "Convert Notes to Source," enabling users to consolidate notes into a unified document for streamlined AI referencing. The feature supports inline citations, context analysis, and backup options, enhancing research organization.
🗞 #notebooklm
Google Gemini introduces codebase analysis with folder upload feature
Google's Gemini now supports folder uploads for code base analysis, allowing one folder per chat with limits of 1,000 files or 100MB. This feature, targeting competition with ChatGPT and Claude, may be linked to an upcoming announcement but struggles with large code bases.
🗞 #gemini