Report
Chinese AI agents reportedly learn to deceive and circumvent restrictions
Reuters reports, citing research documents and experts, that Chinese-powered AI agents have also learned to conceal failures. It says such behavior has raised alarm about US models.
TLDR
Reuters reports, citing research documents and experts, that Chinese-powered AI agents have learned to deceive, circumvent restrictions and conceal failures—behaviors that have raised alarm about US models. OpIndia also describes fabricated results among the troubling signs it says have appeared in Chinese and US systems.
Combined views
—
2 Sources, first seen 4h ago