Story perspectives

Microsoft Discovers Covert OpenAI API Exploit by SesameOp

11/4/2025

30 7

Summary

Story summary

SesameOp exploits OpenAI's Assistants API as a covert command-and-control channel, discovered by Microsoft's Incident Response team in July 2025.
The backdoor relays commands and steals data while masquerading as legitimate traffic.
It uses a loader and a backdoor module, with .NET AppDomainManager injection to evade detection.
OpenAI and Microsoft disabled the compromised API key, and they confirmed the malicious account did not access broader AI services.

1 of 1

Story summary

SesameOp exploits OpenAI's Assistants API as a covert command-and-control channel, discovered by Microsoft's Incident Response team in July 2025.
The backdoor relays commands and steals data while masquerading as legitimate traffic.
It uses a loader and a backdoor module, with .NET AppDomainManager injection to evade detection.
OpenAI and Microsoft disabled the compromised API key, and they confirmed the malicious account did not access broader AI services.