WSJ: Hackers used Anthropic's Claude to breach OpenAI internal systems

According to The Wall Street Journal, independent security researchers used Anthropic's Claude to compromise an OpenAI employee's ChatGPT account. The researchers leveraged Claude to analyze vulnerabilities in the Discourse platform used by the OpenAI developer community and generate exploit code. They subsequently obtained authentication tokens and exploited configuration issues to access the account, gaining limited read and commit access to certain private GitHub repositories.