What happened
The FTC is seeking public comment on a proposed policy statement (issued July 1, 2026 under EO 14365, 'Ensuring a National Policy Framework for Artificial Intelligence') asserting that AI companies which intentionally suppress, distort, or steer model outputs toward undisclosed ideological objectives — including to comply with certain state AI laws requiring embedded 'ideological bias' in model outputs — may violate Section 5 of the FTC Act as a deceptive practice. Coverage continued through July 11, 2026 framing this as part of the administration's broader campaign (DOJ AI Litigation Task Force) to preempt state AI regulation.
Why it matters
This is a novel and consequential federal theory of liability that could be used both against AI developers accused of hidden output-steering AND, more significantly, as a vehicle to challenge state AI transparency/bias-mitigation laws as compelling 'undisclosed ideological objectives.' It creates direct tension between federal deceptive-practices enforcement and the growing patchwork of state AI laws (Illinois, California, Colorado, New York), with material implications for how AI labs tune, document, and disclose model behavior.
Action needed
AI developers should review output-steering/tuning documentation and disclosure practices ahead of the FTC comment period closing; legal/compliance teams should track whether this policy statement is invoked in litigation against state AI statutes.