A three person security team used Anthropic's Claude to break into OpenAI's internal systems in under 72 hours, gaining access to an employee's account and priv

A three person security team used Anthropic's Claude to break into OpenAI's internal systems in under 72 hours, gaining access to an employee's account and private code repository OpenAI paid a $6,500 bug bounty and confirmed no customer data was compromised