4 AI drops worth watching: May 9
OpenAI: GPT-5.5 and GPT-5.5-Cyber for verified defenders
On May 7, OpenAI expanded its Trusted Access for Cyber program with GPT-5.5 and GPT-5.5-Cyber, giving verified defenders access for vulnerability research and critical infrastructure work.
GPT-5.5-Cyber sits behind a gated access model, not a public release.
DeepMind: AlphaEvolve in production across Google’s systems
Google DeepMind detailed AlphaEvolve’s production impact on May 7: a Gemini-powered coding agent that improved Spanner write amplification by 20%, lifted feasible power-flow solutions from 14% to 88%, and generated quantum circuits with 10× lower error than conventional baselines.
External adopters cited in the post: Klarna doubled transformer training speed, FM Logistic improved routing by 10.4%.
Anthropic: higher Claude usage limits, SpaceX compute deal
On May 6, Anthropic announced higher Claude usage limits, backed by a new compute partnership with SpaceX granting access to the Colossus 1 data center: over 300 megawatts and 220,000+ NVIDIA GPUs within the month.
For builders: Claude Code’s 5-hour rate limits doubled for Pro, Max, Team, and Enterprise plans; peak-hours throttling removed for Pro and Max; API rate limits raised for Opus models.
Tilde.run: transactional agent sandbox
Tilde.run launched on May 6. The sandbox turns every agent run into a transaction you can roll back, with a versioned filesystem that mounts GitHub, S3, and Drive into one composable environment.
Free to start, with a private preview for teams running autonomous agents against production data.
More drops at dropwatch.ai. Want them in your inbox? Subscribe below.