The White House’s new approach to advanced artificial intelligence puts industry self-monitoring at the center of safety policy. At a recent gathering, major AI companies signed a voluntary agreement of just 308 words, while the administration has moved away from federal requirements to share certain safety-test results. Technology founders welcomed the lighter regulatory approach, arguing it will accelerate competition and innovation. Former safety researchers and consumer advocates question whether companies can credibly investigate risks tied to their own products without independent oversight. The unresolved issue is accountability: a voluntary pledge may outline intentions, but it does not establish enforceable consequences when systems cause harm. Associated Press examined the pact and the response from Silicon Valley.