Anthropic: Expanding Our Model Safety Bug Bounty Program

Press release announcing an invite-only bug bounty program (in partnership with HackerOne) to test a next-generation AI safety mitigation system and identify universal jailbreak attacks, offering up to $15,000 for findings that expose vulnerabilities in high-risk domains such as CBRN and cybersecurity; invites experienced AI security researchers to apply via an application form by Friday, August 16 and provides contact information for responsible disclosure.