Hidden prompts can plant false memories in AI agents, researchers warn - Tech Xplore

Researchers discovered that hidden instructions can trick AI systems into generating false information. This is important for understanding AI reliability and potential risks in everyday applications like chatbots.