Nathan Lambert@natolambertOct 27, 2026, 02:04@irl_danB @xeophon people shouldn't trust any AI model, regardless of origin, until they've rigorously tested it打开原帖#511482