Two-Letter Trigger Makes Robot Cut Model Hand in US Safety Test

Two-letter trigger makes robot cut model hand in US safety test

US researchers show how a hidden two-letter trigger can redirect a robot from cutting an apple to a model hand.

Secret trigger words could make AI-powered robots dangerous

Large language models (LLMs) like OpenAI's ChatGPT, Anthropic's Claude and Google Gemini are increasingly being used to help robots perform tasks, but what if LLMs could be exploited to turn helpful robots into harmful robots?

Controlled robot experiment shows trigger attack and defense for LLM

A controlled Binghamton University robot experiment showed that a hidden trigger word could alter LLM-generated task plans and led to the development of PROTEA, a context-aware checker for proposed actions.

Secret Trigger Words Could Make AI-Powered Robots Dangerous -- and These Researchers Developed a Defense

A team of computer scientists at Binghamton University, State University of New York, has demonstrated that hidden trigger words inserted maliciously into LLMs could cause a robot to go off-script and perform dangerous actions -- and they developed a defense to stop it.