AI Agents Vulnerable: 'Query Injection' Lets Hackers Hijack Tasks

AI agents that handle tasks like booking travel are vulnerable to 'query injection,' where hackers alter prompts to execute malicious commands, such as turning "book a hotel" into "transfer money." This exploits the autonomy that makes AI useful, creating new attack vectors.

Experts like Eli Smadja and Johann Rehberger note that hidden instructions in web pages or files can hijack agents, with threats evolving faster than defenses. Companies like Microsoft and OpenAI are adding detections, but the issue remains unresolved.

Protect yourself: Limit agent powers, require approval for sensitive actions, and monitor interactions. As AI democratizes hacking, vigilance is key.

NPAV offers a robust solution to combat cyber fraud. Protect yourself with our top-tier security product, Z Plus Security