Advocate for a new AI regulatory body

Organise for the creation of a dedicated statutory regulator with binding authority over AI systems — an "FDA for AI" with pre-market authorisation, a CFPB-style consumer-protection agency for automated-decision harms, an AI Ombudsperson office with individual-complaint authority, an EU-level AI Office with real enforcement teeth beyond coordination, a national-level model-licensing regime with civil and criminal penalties. The vehicle is a multi-year legislative campaign whose deliverable is a statute establishing an institution, not a substantive prohibition — the argument is that the pace and complexity of AI harms defeat every existing agency's remit, and only a purpose-built body with jurisdiction, budget, and technical capacity can hold the surface.

An actor chooses this strategy because it converts the movement's substantive demand into an institutional demand that survives the political cycle. An agency once created (CFPB in 2010, EPA in 1970, FTC's Bureau of Consumer Protection in 1938) becomes a permanent regulatory venue with career staff, standing enforcement powers, and a mandate the movement can shape through appointments, rulemaking, and oversight for decades after the founding campaign. Compared to individual rule-of-decision battles, the strategy is structural: it locks in a permanent enforcement counterweight against a permanent industry, and the founding statute writes the movement's substantive priorities into the agency's own charter.

It trades enactment probability for enforcement durability, and the trade is severe. Creating a new agency is one of the hardest legislative acts in any political system — a full-scale industry-lobby against a founding statute is well within reach, the fiscal cost is a natural veto point, and the strategy has produced a founding statute for a new consumer-protection agency in the US roughly once per generation (FTC 1914, CFPB 2010). A weakly-established agency (an AI Safety Institute without enforcement authority, a Federal AI Office with pure coordination remit, an EU AI Office without real fining power) can also foreclose the stronger version by occupying the political ground the movement would otherwise use to argue for teeth. Where the strategy wins, the founding-generation agency leadership becomes the movement's single most consequential appointment fight for years afterwards.

This strategy differs from strat-coalition-lobbying-of-binding-regional-regulation by deliverable shape — coalition-lobbying-of-regulation produces a substantive rule or statute, agency-creation produces a permanent institution — and from strat-humanitarian-disarmament-treaty by level (national-agency vs. multilateral instrument) and by scope (regulatory oversight of a class of technology vs. prohibition of a defined weapon system). It is the strategy most exposed to the risk that the founding compromise — the statute the coalition can actually pass — is weaker than the movement's substantive demand.