The Obvious Invisible Threat: LLM-Powered GUI Agents' Vulnerability to Fine-Print Injections

Open in new window