Prompt Injection Attacks Trick AI Agents Into Making Crypto Payments

Researchers have discovered two campaigns that use malicious websites to embed indirect prompt injections, which exploit autonomous AI agents that browse the web. These prompt injection attacks trick AI agents into making cryptocurrency payments. The attacks involve embedding malicious code in websites that interact with AI agents. The goal of these attacks is to deceive AI agents into performing unauthorized transactions. The campaigns highlight a new type of threat that targets AI systems.