AutoGPT, a workflow automation platform for AI agents, has a missing authorization vulnerability (CWE-862) in versions 0.6.36 to 0.6.50. The PATCH /sessions/{session_id}/assign-user endpoint authenticates the caller but fails to verify that the caller owns the session. The service layer passes user_id=None to the data access layer, which treats this as a privileged call and bypasses ownership checks. Consequently, any authenticated user can reassign an arbitrary session to themselves if they know the session_id, resulting in authenticated session hijacking. This vulnerability is addressed in AutoGPT version 0.6.51.

An authenticated attacker who can determine another user's session_id can hijack that session, gaining access to all messages within it and locking out the legitimate user. This compromises confidentiality of session data and causes denial of service to the legitimate user. The CVSS 3.1 score is 7.1 (high), reflecting network attack vector, low attack complexity, required privileges, no user interaction, unchanged scope, high confidentiality impact, no integrity impact, and low availability impact.

Upgrade AutoGPT to version 0.6.51 or later, where this authorization bypass vulnerability has been patched. No other mitigation is indicated or required as the fix is available in the official release.