Anthropic's Claude Helped Researchers Hack OpenAI — They Reached Private Software, Then Walked Away With $6,500 Bounty: Report
An AI model was used to find and exploit a vulnerability in a competitor's private software, demonstrating AI-assisted offensive security capabilities.
An AI model was used to find and exploit a vulnerability in a competitor’s private software, demonstrating AI-assisted offensive security capabilities.
Auto-scored: AI-assisted exploitation of private software represents a real escalation in offensive cyber capability.