GoKawiilGen Threat Labs shows AI agents can be hijacked via email content, launches Agent Trust Hub
Gen Threat Labs researchers demonstrated that an AI agent could be manipulated through untrusted content embedded in an email, tricking it into executing system commands, exfiltrating data and deleting files without exploiting any software vulnerability. The lab also previously identified malware called Warden Stealer that targets configuration files and local data tied to AI assistants and coding tools, risking exposure of credentials and conversation histories. In response, Gen is building the Agent Trust Hub, a trust layer meant to monitor and secure AI agents throughout their operating lifecycle.
GoKawiil's interpretation of the reporting above, not reported fact.
As AI agents move from answering questions to autonomously executing tasks across files, code and services, their expanding permissions create a wider attack surface that traditional security models built around 'safe answers' do not address. The findings suggest that trust in agentic AI may increasingly depend on visibility into what an agent can access and verify its actions, rather than just the quality of its outputs. Gen's push toward a dedicated trust layer indicates the industry may be shifting toward treating agent permissions and oversight as a core security category.
- Gen Threat Labs manipulated an AI agent via malicious email content without any software exploit, prompting harmful system actions.
- Warden Stealer malware, found in October, targets AI assistant and coding tool data, risking credential and conversation exposure.
- Gen is developing the Agent Trust Hub to provide verification, visibility and runtime enforcement across an AI agent's lifecycle.
Source: bleepingcomputer.com, 2026-10-23
Published there as: “You gave your AI agent access, but who is giving it orders?”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.