The AI Standards Bill Moves the Platform-Power Fight to Washington

A bipartisan House discussion draft called the Great American AI Act would create a federal framework for artificial-intelligence governance. FedScoop reported that the draft, led by Reps. Lori Trahan and Jay Obernolte, would authorize $100 million per fiscal year from 2027 through 2029 for a Center for AI Standards and Innovation inside the Commerce Department. It would develop voluntary guidelines, best practices, security standards, evaluations, and monitoring of AI progress, while also calling for accountability in government AI adoption. BankInfoSecurity reported that a related House draft would impose safety mandates and independent audits on powerful AI developers, with the fight over a three-year block on state AI laws already dominating the response.

The AI regulation fight has entered its second stage. The first stage was performative: hearings, warnings, executive orders, open letters, industry tours, and everyone pretending the question was whether Washington should do anything at all. That question is over. The real fight now is who writes the rules, who gets protected by them, and who gets preempted before they can act. A bipartisan framework sounds reassuring. A standards center sounds technical. Voluntary guidelines sound harmless. But in platform markets, standards are power. They decide which firms can afford compliance, which startups get buried under process, which state regulators are told to stand down, and which government agencies become permanent customers of the same companies they are supposed to oversee. Jordi’s lens matters here because the incentive map is clearer than the rhetoric. Big AI labs want national rules because fifty state fights are messy. States want room to move because they are closer to consumer harms, labor disruption, deepfake abuse, and local infrastructure costs. Federal officials want authority because AI is now national-security, economic-policy, and bureaucratic-modernization theater all at once. Citizens want something simpler: systems that do not quietly rewrite work, privacy, speech, energy demand, and public services without anyone accountable when things break. The phrase ‘AI standards’ can hide a lot. Standards can be genuine safety plumbing. They can also become a moat. Independent audits can expose risk, or they can become a credentialing market dominated by the same ecosystem. A federal center can bring competence, or it can become a revolving-door checkpoint. The $100 million authorization is not huge by Washington standards, but it signals the institutionalization of AI governance. Once the bureau exists, the next fights will be over model access, liability, procurement, state preemption, and which companies become too embedded to discipline. This is not an argument against regulation. The opposite is true. AI needs rules because platform incentives do not naturally produce public accountability. But regulation that arrives as a bargain between Washington and dominant labs can make the public feel protected while locking in the incumbents. The better test is simple: does the framework increase accountability downward to citizens, workers, patients, schools, and states? Or does it mainly increase coordination upward between federal agencies and the largest AI firms? If it is the latter, the public will get governance theater while platform power becomes federal infrastructure.