AI Security GuruWednesday, November 11, 2026 brief
Microsoft researchers demonstrate hidden backdoor technique for AI agents
Help Net Security
This story is part of the morning brief Railway breach turned stolen customer AI keys into $740,000 of fraud, and NEAR AI loses an inference server to an SGLang RCE. The reporting lives at the original source.
Railway's full report shows an attacker hunting AI provider keys, NEAR AI says confidential inference didn't stop an SGLang exploit, and Microsoft shows a 10-neuron backdoor for agent models.