Meta has launched Muse, a personal AI agent that requests access to user email, calendars, payments, health services and other sensitive data. The release represents Meta's most significant consumer AI initiative and tests whether users will grant AI agents extensive data permissions.
Source: TechCrunch
Hackers have been accessing Claude user accounts and consuming API tokens without authorization, prompting Anthropic to issue a security warning to its subscriber base. The breach was identified when a user detected unexpected token consumption on his account. Source: TechCrunch
Cognition, an AI coding startup, reached a $48 billion valuation in its latest funding round, indicating investor confidence in the AI coding market. The valuation surpasses Cursor's previous valuation before its acquisition by SpaceX, suggesting multiple players can achieve significant scale in the sector. Source: TechCrunch
An NYU mathematician alleges OpenAI engaged in questionable tactics regarding a career-defining mathematics problem with a $1 million bounty. The dispute concerns the Navier-Stokes existence and smoothness problem, one of mathematics' most difficult unsolved challenges. Source: TechCrunch
An AI-generated approach to the Navier-Stokes Millennium Prize Problem has been shared, including both a writeup and a formal proof written in the Lean proof assistant. This represents an application of AI to theoretical mathematics and formal verification rather than a solved prize problem. Source: OpenAI