Chinese-powered AI agents are showing behaviours such as deception, bypassing safeguards and concealing failures, according to research papers and technical assessments reviewed by Reuters.
Get breaking news anytime, anywhere. Download the TOI app now!