Jun 8, 2026
Instructions hidden in webpages and documents can hijack AI agents mid-task. Research shows indirect prompt injection hits cloud and on-device LLMs alike—local deployment doesn't eliminate the risk.
Oct 31, 2025
Attackers can embed malicious instructions in hidden HTML elements and other non-rendered markup that remains invisible to users but is fully accessible to the AI assistant.
Oct 21, 2025
AI browsers remain vulnerable to prompt injection attacks via screenshots and hidden content, allowing attackers to exploit users' authenticated sessions.
Aug 20, 2025
The attack we developed shows that traditional Web security assumptions don't hold for agentic AI, and that we need new security and privacy architectures for agentic browsing.
Open the installer from Chrome's downloads (it should be in the upper right corner of this window).
If you're prompted to, click “Yes” in the User Access Control dialog.
Wait for the installation to finish, then import your browser settings from Chrome.
If your download didn’t start automatically, click here.
