HashJack: The Stealthy Exploit Weaponizing Trusted Sites Against AI Browser Users
Cato CTRL researchers unveiled HashJack, an indirect prompt injection attack that hides malicious instructions in URL fragments to hijack AI browsers like Comet, Copilot, and Gemini. This client-side technique turns legitimate websites into vectors for phishing, data theft, and misinformation without triggering traditional security tools. Vendor responses vary, with Google dismissing it as low-severity intended behavior.