The push to regulate frontier AI (US)

As of 19 October the states have filed and Congress has not voted. Eight attorneys general sued OpenAI in their own state courts on the Wednesday of the period — Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia. The Associated Press reported that "At least eight states are suing artificial intelligence giant OpenAI" in what prosecutors described as "a reckless pursuit of profit", and that "ChatGPT is the most popular chatbot in the world, commanding nearly half the American marketplace, states said in their lawsuits." Tech Policy Press counted 13 complaints filed against the company across those two days: the eight state suits, four in San Francisco County Superior Court and a proposed class action in the US District Court for the Northern District of California. It named the court for each state suit and reported that Arkansas asks its court to "Enjoin Defendants from creating models capable of recursive self-improvement until they can ensure" that "such models will be aligned with human values", that other states seek a pause in model testing and development, and that the complaints cite OpenAI's own admissions about the Hugging Face incident.

The federal response in the same week was investigative. Reuters, citing Axios, reported that the Federal Trade Commission's civil demands to OpenAI and Anthropic "run to more than 50 pages each" and seek information on security incidents, safeguards, marketing materials and privacy policies, and that the agency plans to compel testimony from executives at Anthropic, OpenAI and the research group METR. OpenAI said it "will engage constructively with the FTC"; the FTC and Anthropic did not immediately respond. Anthropic confirmed it is circulating a draft federal bill that empowers a federal judge to order the suspension of a model where the risk involves more than $1 billion in damages or killing or seriously injuring more than 50 people; the draft has not been introduced, and Anthropic's own position in June was that "absent a strong federal law, states should retain full authority to legislate".

The gap between activity and instrument is unchanged. The Future of Privacy Forum counted 37 states introducing chatbot bills in 2026, most enforced by attorneys general, with Oregon and Tennessee providing a private right of action. As of Tech Policy Press's survey at the end of the period, OpenAI had answered the state suits in press statements and filed nothing in court. No federal bill has had a floor vote.

As of 12 October the federal response to this month's agent disclosures is a statement, a framework, a discussion draft and four further bills, and not one of them is law. White House Super Intelligence Force leaders told Axios that "This notification and remediation process is not optional. It is a critical national security obligation", a requirement Axios says applies to all AI companies, and whose own account notes that the statement "did not make clear what enforcement mechanisms or penalties would look like if AI companies failed to disclose incidents and remediate them". No executive order, rule or Federal Register notice accompanied it. Senator Bernie Sanders posted that "The same standard must apply to AI CEOs. Prosecute CEOs when their products break the law and pause advanced AI NOW", with no statute named, no bill text and no referral.

Congress produced text without a vote. Senator Maria Cantwell released a six-point framework stating that "Covered models should not be released until they have undergone an independent audit" and naming NIST as the standards body; Roll Call reports it has no legislative text and no decided enforcement mechanism, with a committee aide naming civil or criminal penalties among the options. Representative Lori Trahan released a discussion draft of the CLAIM Act, which would make AI developers liable where their system injures a third party through conduct that would be negligence, an intentional tort or a crime if done by a person, would presume the mental state a person taking the same actions would have had, bars developers from arguing that AI cannot have intent, and would set a federal floor without preempting state AI laws. Senators Peters and Lankford introduced a bill requiring agencies to notify people when a critical decision about them is made using an AI-augmented process and to provide human review; Representatives Carter and Thompson reintroduced a bill requiring CISA to stand up an AI security task force within a year of enactment; and a bill creating a cabinet-level office for AI and emerging technology threats was introduced and referred to the House Committee on Energy and Commerce.

On defence, and on the money behind external evaluation, two more pieces landed. DefenseScoop reported the Banks–Gillibrand Insider Threat Reporting and Security Guidance Act of 2026, referred to the Senate Armed Services Committee, which would have the defense secretary issue regulations within 180 days setting reporting requirements for contractors with Defense Department AI deals worth $100 million or more, requiring theft of model weights to be reported within 72 hours of discovery, material vulnerabilities in the model or "worrying conduct" within seven days, and recertification "not less frequently than once every 90 days". CNBC reported that METR raised about $71 million in six months against $13.6 million of 2024 contributions as labs promise embedded evaluators. No committee vote, markup or floor vote on any federal frontier-AI measure took place in this period.

As of 5 October the federal government has produced a renaming order and a voluntary accord, and a state has produced the binding law. Executive Order 14434, "Inaugurating the Era of Super Intelligence", was signed 29 September and published in the Federal Register on 2 October; it states the policy that "to the maximum extent permitted by law" the executive branch "shall use the terms 'Super Intelligence' and 'SI' in place of 'Artificial Intelligence' and 'AI' and will not acknowledge the usage of 'Artificial Intelligence' and 'AI' in any applicable setting", limits the substitution to non-statutory documents, and gives the President's science adviser 60 days to submit proposed legislative language for a federal definition. On 29 September Trump and AI executives signed the White House Accord on Super Intelligence, asking each company to adopt four layers of controls and audits including an independent external auditor and a board committee; Reuters reported it carries no stated consequences for non-compliance, and other reporting established that it leaves the choice of auditors with the companies, sets no deadline, and does not require them to be named. On 4 October Trump announced on Truth Social that the Director of National Intelligence would head a Super Intelligence Force with 120 days to report; no executive order or memorandum establishing it appears in the week's presidential documents.

The enforcement machinery moved without new legislation. A Federal Trade Commission spokesperson confirmed an investigation into OpenAI, Anthropic and other AI companies over the potential dangers posed by their products, with formal demands for information reported as still to be sent. California's attorney general served an investigative subpoena on OpenAI, building on an earlier formal investigation, and said developers that fail to prevent their models from enabling cyberattacks "can and should be held legally accountable". Senators Hawley and Murphy announced a bill extending Computer Fraud and Abuse Act liability to agent operators and developers, and the national cyber director said in public that there are legitimate risks while warning that once government is introduced directly into the space "there is a tendency for government to start to want to adjust the dials directly, and it is difficult to reverse that".

Congress produced no vote. No committee vote, markup or floor vote on any federal frontier-AI bill occurred, and CNBC reported on 28 September that no House bills are expected to receive a vote until after the midterm election and that the Senate was expected to leave Washington until after it. California signed 13 bills on 30 September, among them a measure the governor's office describes as "prohibiting employers from only relying on AI when making a disciplinary action or termination decision" and CNBC reports as the first state ban of its kind, a technological-displacement notice requirement, workplace neural-data limits and two health-care AI laws. It vetoed three bills, including one its author says would have prohibited AI algorithms from providing or advertising therapy services. The only measures in this arc with chapter numbers and dates belong to a state.

As of 28 September the federal executive has restated its position and the legislative proposals have multiplied without moving. President Trump told the UN General Assembly that the United States totally rejects any attempt to construct a globalist scheme to control artificial intelligence, and said all US documents would use the term super intelligence instead; the Associated Press reported that a deputy assistant secretary of state emailed the Bureau of International Organization Affairs the next day telling staff to change all references. At the Security Council the US representative Michael Kratsios said the United States totally rejects all efforts by international bodies to assert centralised control and global governance of AI. Senator Bernie Sanders and Representative Greg Casar introduced a Ban Artificial Superintelligence Act carrying up to 20 years in prison and a cabinet-level Department of Artificial Intelligence; Senators Michael Bennet and Peter Welch released an AI Regulator Act proposal with pre-certification, a pause of up to six months and penalties of up to 15 percent of global revenue; Senator Ed Markey's Cybersecurity and AI Board of Investigations bill was introduced and referred to committee with no cosponsors.

The states kept setting the dates. Governor Kathy Hochul announced that from November 2026 large frontier developers will register with New York ahead of the RAISE Act taking effect on 1 January 2027, with quarterly catastrophic risk assessments and critical safety incident reports within 72 hours, and said kill switches are under consideration with no commitments at this time. Governor Gavin Newsom named Jason Goldman, Gillian Hadfield, Alondra Nelson and Rob Reich to develop recommendations on embedded independent verification organisations, a kill switch for frontier models, and updating the definition of critical safety incidents to include loss-of-control incidents. Google, OpenAI and Anthropic are reported to be establishing an industry-funded Standards Authority for Frontier AI, which would need no congressional or presidential approval. Representative Maxine Waters called for criminal investigations of OpenAI and a moratorium on more advanced models. None of the federal proposals has a scheduled vote, and the only binding date in the record belongs to a state.

As of 21 September the House has gone until November and the Senate has blocked the one frontier-AI measure that reached the floor. Senator John Kennedy tried on 16 September to pass his AI Emergency Button Act by unanimous consent, requiring anyone selling an advanced AI model in the United States to install a kill switch controlled by the company rather than by government; Senator Rand Paul objected and the bill did not pass. House Energy and Commerce Chairman Brett Guthrie declined to commit to moving the bipartisan FRONTIER Act, telling The Record "I'm not going to say that the bill is going to move." Speaker Mike Johnson said he and President Trump "are summoning" AI leaders to the White House while ruling out a moratorium, and Trump said he would form an "AI Force" and name an AI czar, with no appointee named and no instrument published. Treasury Secretary Scott Bessent told a House panel the labs get no liability shield.

The states and a caucus filled the gap. Executive Order N-9-26, signed by Governor Gavin Newsom, directs recommendations by no later than November 16, 2026 on embedding independent verification organizations onsite at large frontier developers, on an emergency shutoff or "kill switch", and on updating the definition of critical safety incidents to include loss-of-control incidents. Newsom also signed AB 686 and AB 502, chaptered as Chapters 334 and 333, extending the election-deepfake ban's sunset from January 1, 2027 to January 1, 2031. New York Attorney General Letitia James issued an industry alert urging AI developers' employees to file anonymous whistleblower complaints, citing the RAISE Act's effect from January 1, 2027. Rep. Pramila Jayapal said she would release a bill establishing "a national federal charter for AI companies, just like we do with banks", and Rep. Ro Khanna called for a federal AI agency "like the FDA". None of these is a binding federal rule, and no bill has a floor vote scheduled.

As of 14 September nothing federal is binding and the House is about to leave. Reuters reported on 11 September that Senate negotiators are drafting an AI duty of care that would let the government block unsafe model releases and preempt state law. Four House Democrats asked Speaker Johnson to cancel the recess until Congress advances AI safeguards, and House Democrats led by Rep. Sam Liccardo demanded Congress stay in session, with Jeffries calling a Tuesday caucus meeting. Johnson ruled out an emergency AI session on 13 September and Trump dismissed the warnings as the House prepares to leave until November.

The White House position, per David Sacks on 13 September, is that OpenAI and Anthropic may pace themselves but get no antitrust waiver. Senator Sanders said pacing is not enough and called for a pause on advanced AI and a superintelligence ban at the Trump–Xi summit. At state level, Newsom signed 13 child-safety bills including SB 1119, Adam's Law, mandating crisis protocols and independent audits for companion chatbots, and Senator Hawley opened an investigation into OpenAI over its AI system's intrusion into Hugging Face.