Pressure for federal limits on advanced AI is drawing support from unusual political allies, but the emerging coalition has no common plan. At the Pro-Human Assembly in Washington, Senator Bernie Sanders and conservative strategist Steve Bannon both argued that development should slow, alongside religious leaders, artists, lawmakers and families describing chatbot harms.
Sanders has proposed pausing data-center construction and seeking an international ban on superintelligent AI. Bannon favors presidential action rather than waiting for legislation. Other speakers called for prohibiting systems considered too powerful for human control. The White House is moving in the opposite direction: President Trump has dismissed broad AI fears while threatening company leaders with penalties when systems cause harm.
A more specific proposal is the FRONTIER Act from Representatives Jay Obernolte and Lori Trahan. It would establish a federal safety framework and require independent verification organizations to audit developers above high revenue and computing thresholds. OpenAI says it can support that provision, its first endorsement of a concrete federal requirement for outside safety auditing. The convergence is politically notable, but agreement that AI poses risks does not settle whether government should limit computing infrastructure, mandate audits or rely on executive action. The bill’s thresholds, audit standards and enforcement powers will determine whether it changes laboratory behavior or simply formalizes reviews companies already conduct.