By Vero SecurityBuilt and funded by Vero Security, the AI Security Engineering platform.Read the charter

Data exfiltration channels

Once an agent is hijacked, the attacker still needs a way to get data out. Any feature that makes an outbound request can serve as that channel, including rendered images, links, diagrams, DNS lookups, and ordinary tool calls.

Read in this order

  1. ArticleCopirate 365 at DEF CON: Plundering in the depths of Microsoft Copilot, CVE-2026-24299 (Embrace The Red)A full 2026 exfiltration chain against an enterprise assistant, with demo videos.
  2. VideoJohann Rehberger - Your Agent Works for Me Now | [un]prompted 2026Demonstrates exploit chains, persistence and data exfiltration. 26 min.
  3. PaperEchoLeak: The First Real-World Zero-Click Prompt Injection Exploit in a Production LLM SystemA zero-click attack on a production assistant, analyzed in technical depth.

Save your progress

We email you a sign-in link. No password.

By signing in you agree to the terms. The privacy policy explains how we use your details.

Join the AI security engineering community

Free. Roadmap updates, chapters near you, and first word when credentials open.

We’ll email you community updates. Unsubscribe anytime. By sending this you agree to the terms. The privacy policy explains how we use your details.