OpenAI said on November 6 that it partnered with Hacktron, an AI security startup, to jointly test high-priority parts of its systems, especially connections to third-party services.

OpenAI said it was expanding Hacktron’s access to frontier models through Daybreak Red, part of a cyber defense initiative that allows trusted external teams to stress-test systems for vulnerabilities.

Hacktron builds AI agents for code review, penetration testing, and red teaming. The company previously raised US$2.9 million to build an AI-powered platform for automated security testing.

Security researchers warn that AI systems connected to external tools can expand attack surfaces through prompt injection, insecure application programming interface integrations, and data exfiltration.

Policy analysis suggests that European Union requirements and the US National Institute of Standards and Technology’s voluntary guidance are pushing organizations toward more formal red-teaming and cybersecurity testing.

The move follows OpenAI’s announcement that researchers from Cathedral will join its next security sprint and that it had increased bug bounty rewards for external researchers.

🔗 Source: OpenAI

Recent OpenAI developments