we_are_coded.by CODE · The world, decoded
БГ

Concept

Daybreak Red

OpenAI's program that gives outside security researchers access to its frontier models to hunt for weaknesses in its own systems.

Updated on 9 November 2026 · Checked on 9 November 2026

Daybreak is the window OpenAI opens to outside security researchers. Through it they get access to the company's frontier models to look for weaknesses in its own systems. The logic is simple: if the models help defense, let them help the people who test it too.

The first public example is Hacktron. On November 6, 2026, OpenAI announced a partnership with the team that builds AI agents for code review, penetration testing, and red teaming. OpenAI's engineers and Hacktron's researchers pick the high-priority areas to test together, with a focus on the seams: how OpenAI's systems connect to third-party services.

That's where it gets interesting. At every link between two systems, nobody holds the whole picture. Hacktron is not a stranger in the house: OpenAI got to know it through its bug bounty program and says it was impressed by the impact of the vulnerabilities its researchers found and reported.

There is no CVE number here. This is not a hole in a product, it's a way to look for holes.

The partnership is not just access. The same week, OpenAI said Cathedral's researchers will work alongside its teams in its next security sprint, and raised the rewards in its vulnerability reporting program.

The question is not who gets in, but what comes out. We'll see what they find.

Primary sources

OpenAI

The visual is generated code-art, no third-party images.