Researchers exploit vulnerabilities to access OpenAI employee accounts
September 18, 2026
A chain of two vulnerabilities allowed researchers to compromise OpenAI employee ChatGPT accounts. The exploit potentially granted access to internal repositories and monorepos, highlighting significant security risks in AI platforms.
HOW THIS AFFECTS YOU
●
builderAudit your use of enterprise AI tools to ensure they do not become vectors for internal data breaches.
●
policyThis underscores the urgent need for stricter security protocols around model access and employee toolkits.