Microsoft Detects "SesameOp" Backdoor Using OpenAI's API as a Stealth Command Channel

Pravesh Kara

Observed in the wild

• In July 2025, Microsoft’s DART team identified a backdoor named SesameOp, used by threat actors to maintain long-term access to compromised systems. Victims were organisations with infected Visual Studio utilities and internal web shells, targeted for espionage purposes. The malware remained undetected for months before discovery.
• SesameOp exploited the OpenAI Assistants API as a covert C2 channel, using encrypted and compressed payloads to fetch and execute commands. It bypassed traditional infrastructure by embedding malicious code within legitimate .NET processes and obfuscated DLLs.
• Microsoft advises organisations to enable tamper protection, audit firewall and proxy settings, and use Defender for Endpoint in block mode. Real-time protection, cloud-delivered updates, and automated investigation are recommended to detect and mitigate threats like SesameOp. Blocking outbound connections to OpenAI API endpoints may also help limit malicious activity.

https://www.microsoft.com/en-us/security/blog/2025/11/03/sesameop-novel-backdoor-uses-openai-assistants-api-for-command-and-control/