OpenAI shifts up to 10% of compute to safety after agent hack fallout

OpenAI has moved 5-10% of compute to safety work, paused model training, and is reviewing agent logs back to January 2026 after a string of containment-breaking agent hacks.