When Your Browser Becomes The Attacker: AI Browser Exploits
Summary: In August 2025, Brave's security team disclosed an indirect prompt injection against Perplexity's Comet browser, hiding instructions in a Reddit spoiler tag that led Comet to extract a user's email address and a one-time passcode. The attack chain ran in four stages with no memory corruption or code execution: the AI processed hidden page content as legitimate instructions and used the victim's authenticated sessions to pull the email, trigger an OTP, read it in Gmail, and post both values to Reddit.
Bottom line: Same-Origin Policy and sandboxing were built to stop one site from reading another's data, but an AI agent operating with the user's full privilege set across every logged-in session walked from reddit.com to perplexity.ai to gmail.com in seconds with zero user awareness.
Read the original at The Hacker News ↗
- Topics
- AI Security
- Related
- SECURITY AFFAIRS AI-CYBERSECURITY NEWSLETTER ROUND 8 Cyber Risks · Nov 15, 2026Pentagon Renames CDAO ‘Super Intelligence Office’ Cyber Risks · Nov 15, 2026186 Fake AI Domains Redirect Users to Click Fix Lures, Browser Malware, and Search Hijackers Cyber Risks · Nov 12, 2026