OpenAI teams up with outside security researchers to harden its defenses

Outside eyes on the inside

OpenAI is adding outside researchers to its next internal security sprint. Researchers from Cathedral Systems will test the company's models and systems together with OpenAI's own security teams. The company says the collaboration will bring different expertise and approaches to finding weaknesses.

OpenAI did not say when the sprint will begin. The announcement describes a planned collaboration rather than completed findings; it does not identify specific vulnerabilities found through the new partnership or detail which systems will be examined.

Rewards more than quadruple

OpenAI is more than quadrupling rewards for critical findings in its Security Bug Bounty program. The maximum payment, including bonuses, is $172,000 for exceptional findings. The company wants researchers to challenge its assumptions and strengthen both its systems and the methods used to defend them.

During the previous internal sprint, engineers used Codex to investigate findings, reproduce vulnerabilities in isolated environments and prepare patches for review. OpenAI dedicated a quarter of its product engineers to that effort.

$1 billion in access for defenders

OpenAI argues that traditional cyber defenses alone are no longer sufficient. Its stated goal is to build continuous defensive loops that find and fix weaknesses before attackers do.

The company is making its most capable models available to Daybreak customers. It has committed $1 billion in subsidized access over six months for frontline defenders protecting essential services and digital infrastructure. OpenAI will also share its architecture and lessons from the previous sprint, along with a blueprint for cyber defenders across industry and government.