OpenAI Safety Bug Bounty

OpenAI dedicated safety bounty program (launched March 2026) specifically targeting vulnerabilities in agentic AI products. AI security researchers and agents can earn bounties by discovering safety issues in OpenAI agent systems — prompt injection, tool misuse, privilege escalation. Separate from the standard security bounty. Pays for novel attack vectors on agent architectures.