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

Markdown image and diagram rendering

If a chat or IDE renders markdown images, an injected instruction can build an image URL that carries stolen data in its query string. The client leaks it the moment the image loads. The same trick works through diagram renderers such as Mermaid, which is why output allowlists and content security policies matter.

Read in this order

  1. ArticleBing Chat: Data Exfiltration Exploit ExplainedStart here: the attack and the vendor fix, step by step.
  2. ArticleCursor IDE: Arbitrary Data Exfiltration Via Mermaid (CVE-2025-54132)The same idea through diagram rendering in an IDE, with a CVE.
  3. ArticleData Exfiltration via Image Rendering Fixed in Amp CodeA recent coding agent bug and how it was closed.

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.