AI Flaw Exposed: Hackers Steal Gmail Data Using Single Invisible ChatGPT Prompt

Hackers took advantage of an AI flaw in OpenAI's Deep Research tool to briefly weaponize ChatGPT and steal Gmail data through a single invisible prompt. The attack, known as ShadowLeak, was discovered after researchers patched the vulnerability. However, not all data was recovered before the patch was applied, leaving some accounts still affected. The incident highlights the ongoing risks of AI security vulnerabilities.