AI Trends — 31 August 2026

OpenAI to Cut Off Cursor After SpaceX Acquisition

OpenAI announced plans to terminate the contract that lets coding platform Cursor use its models, with access ending November 12, 2026. The move follows Cursor's acquisition by SpaceX; OpenAI says it cannot be confident its technology would continue to be used under the original contractual terms once Cursor changes hands. The decision highlights how model access is becoming a lever in the competitive and geopolitical landscape around AI tooling.

Music Publishers Sue Anthropic Over Training Data

Sony Music Publishing, Warner Chappell Music, and other major publishers filed suit against Anthropic in California on August 28, alleging the company used torrenting, scraping, and bulk downloads of copyrighted songs to train Claude models. The case adds to a growing pile of copyright litigation against frontier AI labs and could shape how courts treat large-scale data collection for model training.

IBM Claims Quantum Advantage on Practical Workload

IBM and University of Chicago researchers reported completing a computation using 70 error-corrected logical qubits in roughly 15 minutes, a task they say leading classical methods could not practically reproduce. If the result holds up to scrutiny, it would mark a notable step toward quantum computing delivering on real-world problems rather than narrow benchmark demonstrations.

Anthropic Responds to Account Security Incident

Anthropic is signing out affected Claude users, clearing saved payment methods, and refunding unauthorized charges after infostealer malware on users' PCs was found siphoning active Claude login sessions and burning through usage limits without authorization. The episode is a reminder that account security threats increasingly target AI service credentials directly, not just the underlying infrastructure.

Market and Governance Watch

Anthropic's run-rate revenue has passed $30 billion, up from roughly $9 billion at the end of 2025, with business customers spending over $1 million annually more than doubling in under two months. Separately, a chatbot accuracy test by NPR and NewsGuard found six major assistants, including ChatGPT, Gemini, and Claude, correctly debunked false narratives about three-quarters of the time, outperforming traditional search. On the policy side, momentum is building around agent governance, with Google's Agent Payments Protocol, NIST's work on agent identity and permissions, and the proposed AI AGENT Act all aimed at creating verifiable, task-bounded authorization for autonomous AI agents.

This newsletter is generated daily, summarizing AI developments from the prior 48 hours.