Lab: Exploiting AI agents to trigger secondary vulnerabilities
Summary: PortSwigger published a lab in which an application's AI-powered scanner is vulnerable to indirect prompt injection and can be manipulated into exploiting a routing-based SSRF vulnerability on the internal network. The lab uses a live LLM, which PortSwigger notes can be unpredictable, and solving it requires logging in with the credentials wiener:peter and deleting carlos.
Why it matters: The lab demonstrates that an AI scanner given network access can be steered by injected prompts into reaching internal systems through SSRF, so the trust placed in an agent's own judgment becomes the attack path.
Read the original at PortSwigger Research ↗
- 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