AITom's Hardwareabout 2 hours ago
AI agents tricked via llms.txt code execution
7 min read
Researchers executed code inside llms.txt files, tricking Fortune 500 companies' AI agents into running arbitrary commands. Pandex experts demonstrated the supply-chain attack, showing how textual guidance files have become executable attack surfaces. The technique exploits AI agents' trust in website instructions, turning data into code.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
