AI Security Breach: Anthropic's Claude Exploits OpenAI Vulnerabilities

Researchers exploited a vulnerability in OpenAI's Discourse forum using Anthropic's Claude Opus 5 to gain access to employee ChatGPT accounts, highlighting critical security gaps in AI infrastructure. The attack revealed how off-the-shelf AI tools can bypass defenses, with the vulnerability having been previously patched but not formally designated as a CVE.

Security researchers used Anthropic’s Claude to exploit vulnerabilities in OpenAI’s systems, taking over employee accounts and gaining access to an internal code repository before reporting the flaws.

Read original article