From the source
OpenAI says it is planning another internal security sprint, bringing outside researchers in to test its systems with its own models.
- The company is partnering with Cathedral, whose researchers will join OpenAI teams in looking for weaknesses and testing defenses, a job OpenAI frames as continuous rather than seasonal.
- OpenAI is also increasing rewards in its Security Bug Bounty program, more than quadrupling the payout for critical findings so researchers challenge its assumptions.
- The announcement builds on a previous sprint in which OpenAI dedicated 25% of its product engineers to finding and fixing weaknesses, with Codex used to investigate findings, reproduce vulnerabilities in isolated environments and prepare patches for review.
- OpenAI says traditional cyber defenses alone are no longer sufficient, pointing to recent industry incidents as evidence that threats emerge quickly.
- It has committed $1 billion in subsidized access over six months for frontline defenders protecting essential services and digital infrastructure, and is sharing its architecture and lessons from the last sprint alongside a blueprint for cyber defenders in industry and government.
