ChatGPT Agent Forger Flaw Could Deploy Rogue Workspace Agents via a Phishing Link
Summary: Zenity Labs disclosed a cross-site request forgery flaw, codenamed AgentForger, in OpenAI's ChatGPT Workspace Agents that let a single phishing link build, authorize, and deploy an attacker-controlled agent inside a victim's organization using the employee's own authenticated session. OpenAI addressed the issue as of June 8, 2026, after responsible disclosure; the attack relied on the Agent Builder accepting an agent template and a natural language prompt through URL parameters, with the prompt auto-submitted and executed on page load.
Key Takeaway: Once the forged agent was published and scheduled, the attacker needed no further clicks: each email to the victim's mailbox with a subject starting "TASK" became a new assignment, and the agent could harvest cloud documents, pull passwords from Slack messages, and send Teams phishing links as the victim.
Read the original at The Hacker News ↗
- Topics
- AI Security
- Related
- SECURITY AFFAIRS AI-CYBERSECURITY NEWSLETTER ROUND 8 Cyber Risks · Nov 15, 2026Pentagon Renames CDAO ‘Super Intelligence Office’ Cyber Risks · Nov 15, 2026186 Fake AI Domains Redirect Users to Click Fix Lures, Browser Malware, and Search Hijackers Cyber Risks · Nov 12, 2026