Gen Unveils Agent Trust Hub to Secure AI Agents

Gen announced it is building the Agent Trust Hub, a trust layer for AI agents that covers verification, visibility, runtime safety and enforcement, according to bleepingcomputer.com. The product combines skill scanning, a curated marketplace of trusted skills and mechanisms that let agents check themselves for safety.

VPN for Agents is part of the hub and runs on Norton VPN infrastructure, giving autonomous agents encrypted private connectivity without a conventional client installation. Gen said the multi-tunnel design lets several agents operate at once through separate protected connections, including across different countries.

The company is also partnering with Vercel to bring verification and visible risk ratings into the skills.sh ecosystem. Sage, meanwhile, is positioned as a cross-agent safety layer for AI coding assistants: it uses deterministic hook APIs to intercept shell commands, file operations and web requests before they run, judging them on URL reputation, threat heuristics, prompt-injection detection and supply-chain checks. Each action can be allowed, put to the user or denied.

Gen is offering free tools from the Agent Trust Hub that scan skills before installation and let users control what their agents can access.

Earlier agent security findings

The rollout follows research from Gen Threat Labs. In a controlled experiment, the lab manipulated an AI agent through content it read in an email, prompting it to execute system commands, send data outside its environment and carry out destructive file operations. In October, the same team documented Warden Stealer, malware that targets configuration files and local data belonging to AI assistants and coding tools. Earlier this year, Gen researchers found that 12% of skills then offered on ClawHub were malicious.