OpenCode says AI models helping defend against an ongoing cyberattack began refusing further assistance mid-incident, with access to Daybreak Blue sub...

OpenCode disclosed that a breach tied to a large-scale distillation attack may have exposed OpenAI, Anthropic and Google API keys belonging to about 1.4% of its Console users, who were told to revoke and replace them. During the incident, co-founder Dax Raad said, the AI models the team used to analyse the attack began refusing requests once safety filters were triggered, and access to OpenAI's Daybreak Blue defender programme improved the response. The episode highlights an asymmetry: defenders can be blocked by the same guardrails that attackers ignore.

• Kim2026-10-18 19:02