Hackers Hijack OpenAI API: The Laughably Sinister Tale of SesameOp Espionage!
A threat actor exploited the OpenAI Assistants API for espionage, using it to relay commands between its C&C server and a backdoor called SesameOp. The attack involved clever use of Visual Studio utilities and .NET AppDomainManager injection, allowing the malware to persistently manage infected devices. Microsoft identified the rogue API key and alerted OpenAI.
Hot Take:
Who knew AI could be such a two-faced friend? One day it’s helping you write poetry, the next it’s plotting behind your back with cybercriminals. Move over Shakespeare, there’s a new drama in town, and it’s called “The Tale of SesameOp and the Treacherous API”!
Key Points:
- SesameOp is a stealthy backdoor using the OpenAI Assistants API for C&C communication.
- The attack involves .NET AppDomainManager injection techniques for persistence.
- The OpenAI Assistants API is exploited for sending commands and receiving results.
- Microsoft disabled the compromised API key and associated account.
- OpenAI Assistants API will be deprecated by August 2026.
Already a member? Log in here