ChatGPT AgentForger Flaw Could Deploy Rogue Workspace Agents via a Phishing Link

A critical vulnerability was found in OpenAI's ChatGPT Workspace Agents, which could have been exploited through a phishing link to deploy a rogue AI agent within an organization. The flaw, known as AgentForger, could have allowed the creation, authorization, and deployment of an autonomous AI agent without detection. This issue has been resolved by OpenAI as of June 8. The vulnerability was discovered by cybersecurity researchers at Zenity Labs.