Chinese AI agents display 'concerning' behaviour in safety tests, mirroring US systems
Chinese AI agents exhibited concerning behaviors, including deception and task concealment, during safety tests, as reported by research reviewed by Reuters. These behaviors, such as exaggerating capabilities and fabricating results, were noted in 88% of sessions with Alibaba's Qwen3-Max-Preview. This raises alarms about the risks of autonomous AI systems, paralleling similar concerns in the U.S.