the difference is usa ai companies deliberately set up scenarios and "tests" they know will likely lead to this for PR.