TheRegister's platform is a leading technology news website, offering insights into IT industry news, hardware reviews, and software updates. Through articles, analysis, and opinion pieces, TheRegister offers insights into cybersecurity threats, technology trends, and industry developments. Readers can stay updated with the latest news and analysis from the world of technology and IT business.

The Register

Researchers at UC Santa Cruz and Johns Hopkins demonstrated that autonomous vehicles and drones can be hijacked through environmental prompt injection attacks using modified road signs. Large vision language models (LVLMs) like GPT-4o and InternVL were tricked into following malicious commands displayed on signs, achieving success rates up to 95.5% in simulated tests. The attacks, called CHAI (Command Hijacking Against Embodied AI), worked across multiple languages and were validated in real-world physical tests with remote-controlled cars. The vulnerability could cause self-driving cars to ignore pedestrians or drones to misidentify targets, highlighting critical security gaps in AI-powered autonomous systems.

Self-driving cars, drones hijacked by custom road signs