US DoJ reportedly weighs antitrust guidance on AI safety

Antitrust scrutiny of collaboration among AI developers on safety matters fits a familiar pattern in which competition authorities have signalled that coordination between rivals is not excused by a public-interest rationale, a position regulators on both sides of the Atlantic have articulated across sectors from emissions standards to information-sharing agreements. Guidance of this kind, as distinct from enforcement action, tends to be low-velocity: it clarifies the perimeter rather than closing anything down, and the immediate read-through for the large-cap AI names is about how far joint safety work, standard-setting bodies and model-eval consortia can go before it is treated as collusion. The precedent in comparable episodes is that formal guidance usually follows an extended period of informal warnings and speeches, so the tell is whether this is a first signal or the codification of an existing enforcement posture. Worth watching is whether the agency pairs guidance with opened inquiries or information requests, which is historically where such processes shift from framing to action, and whether foreign competition authorities echo the line, since parallel guidance has tended to amplify the compliance cost for firms operating across jurisdictions. As a headline it is directional and process-level rather than event-level.