US DOJ Weighing Antitrust Guidance on AI Safety, Official Says

The signal here is jurisdictional: AI safety is being framed as a competition question, not purely a technical or national-security one. That matters because safety practices agreed among model developers — shared evaluations, disclosure norms, compute thresholds — could be read as coordination among rivals, putting the largest labs in the antitrust crosshairs rather than only the safety regulators. The absence of a timeline or formal proposal means this is intent, not policy. Whether the DOJ actually publishes guidance, and how narrowly it defines the conduct at issue, is the open question.