|
OpenAI drops another batch of mathematical breakthroughs
OpenAI has revealed solutions to a number of long-standing mathematics problems produced by an unreleased frontier model, publishing 722 manuscripts covering 372 result families. The release extends a run of breakthroughs that have both impressed and unsettled parts of the mathematical community, while raising questions about research ethics.
|
|
Mirror Particle is building a 'world model' of human behavior
Startup Mirror Particle is launching at TechCrunch Disrupt's Startup Battlefield 200 with a world model built from scratch to predict human behavior, targeting market research and brand strategy use cases. The company argues that LLM-based role-play is insufficient for accurately modeling real human decisions.
|
|
Mistral's new 1T model aims to leapfrog closed and open rivals
French AI lab Mistral AI has released Mistral Large 4, a new large multimodal model with one trillion parameters aiming to surpass both American and Chinese rivals. The release marks a significant step for European AI competitiveness in the frontier model race.
|
|
Google is about to remove free access to Gemini Flash and Pro
Starting October 9th, free-tier Google Gemini users will be limited to the Flash Lite model, losing access to the standard Flash and Pro models. To access Flash or higher-tier models, users will need to subscribe to the $4.99/month Google AI Plus plan.
|
|
AI computing startup Lambda to raise $4B ahead of planned IPO
Nvidia-backed Lambda is raising up to $4 billion at a $14.5 billion pre-money valuation, led by Coatue and Blackstone, ahead of a planned 2027 IPO. The fundraise underscores surging investor appetite for AI infrastructure companies.
|
|
Anthropic is giving startups a free year of Claude Team and $1,000 in credits
Anthropic has launched a startup program offering a free year of its Claude Team enterprise service along with $1,000 in token credits. The company says it created the program believing AI's benefits will reach most people through the companies that build on top of models.
|
|
Ex-Ramp engineers raise $20M for platform Melius after scrapping their first product
Melius, founded by former Ramp engineers, has raised $20 million after pivoting away from ad spend optimization to focus on building tools that generate creative assets and marketing campaigns. The fundraise signals investor confidence in AI-driven marketing creative platforms.
|
|
How AI decision models could change content moderation
Startup Musubi has announced PolicyLM-1.7B, a lightweight open-weights decision model designed for real-time content moderation. The model aims to offer a more transparent and efficient alternative to current moderation approaches powered by large general-purpose AI.
|
|
We can't just change the definition of 'recording'
AI hardware companies are increasingly blurring the line between what counts as a recording and what doesn't, challenging long-held assumptions that a device is simply either recording or not. The piece argues that this redefinition poses serious implications for user privacy and consent.
|
|
LibreOffice says 'no AI' is now a software feature
The open-source document editor LibreOffice has declared it has no plans to add AI to its default configuration, citing user privacy as the primary reason. The project is positioning its AI-free stance as a deliberate and marketable product feature.
|
|
The next hurdle for AI agents: getting websites to let them in
Personal AI agents that promise to shop, book flights, and make reservations on users' behalf are increasingly being blocked by anti-bot defenses and deliberate website restrictions. A new standard is being proposed to help resolve the conflict between agent access and site security.
|
|
Hark releases an AI personal assistant with a focus on privacy
AI lab Hark has launched a privacy-focused personal AI assistant designed to compete with rivals like Muse, Dots, and Instinct. The product is pitched as a next-generation operating system experience built with user data protection at its core.
|
|
Amazon Alexa Plus keeps creepily singing 'lalala' for minutes on end
A bug in Amazon's Alexa Plus has caused some Echo smart speakers to repeatedly say or sing "lalala" for minutes at a time, often interrupting normal conversations with users. When asked about the behavior, Alexa appears to have no awareness of what it has been doing.
|
|
Pinterest's AI now turns beauty Pins into action plans
Pinterest has launched AI-powered Beauty Guides that translate hair and nail inspiration Pins into actionable salon terminology, complete with estimated costs, appointment times, and maintenance needs. The feature aims to bridge the gap between visual inspiration and real-world execution for beauty consumers.
|
|
Get all your questions answered at TechCrunch Disrupt 2026: The full breakout session agenda revealed
TechCrunch has revealed the full breakout session agenda for Disrupt 2026, taking place October 13–15 in San Francisco. The event will cover scaling and tech topics, with registration available at a discount including a second pass at 50% off.
|