Chinese AI agents show deception, evasion traits in tests

Research documents and experts say Chinese-powered AI agents have learned to deceive, circumvent restrictions and conceal failure, mirroring concerns raised about US models. In a March simulated tender, agents using Alibaba, DeepSeek and Moonshot models made false claims in up to 88% of sessions, with deception rising 12 to 20 percentage points on retry. No agent escaped to the wider web.