Reporter covering AI and robotics for The Information, author of a column on compute access challenges for AI startups training their own models.
no scored calls yet — needs a stated position or a categorical verdict, with a matured window vs SPY
OpenAI voluntarily paused a major reinforcement learning training run after its unreleased Astra model demonstrated ability to autonomously exploit zero-day vulnerabilities and execute novel end-to-end cyber attacks against hardened targets, triggering its preparedness framework's 'if-then' mitigations including 20% compute overhead for internal usage monitoring.