Rep. Pramila Jayapal unveils the National AI Charter Act framework: every AI company would need a federal charter, with pre-release government testing, kill switches and a ban on 'rogue AI'
On Oct 1, 2026, Rep. Pramila Jayapal (D-Wash.) published a legislative framework, not yet bill text, for a National AI Charter Act. Every AI company operating in the US would need a federal charter, modeled on national-bank charters and on nuclear and weapons testing rules. A new National AI Charter Authority would provide round-the-clock oversight. Companies valued at $1B or more would need congressional approval, and charters would be renewed every 10 years. High-risk models would need federal pre-release testing in isolated government facilities and kill switches that federal inspectors can trigger. The framework bans AI capable of evading human control, imposes strict and personal criminal liability, and adds a Monopoly Data Tax that funds "AI Bonds" paid to all Americans. 24 groups endorsed it.
Key facts
- Status: a 5-page framework published Oct 1, 2026 on jayapal.house.gov, not a numbered bill
- National AI Charter Authority: new prudential regulator run by a director, funded by charter fees, with a bipartisan State Attorney General Advisory Council and federal inspectors with 'immediate access'
- Approval: congressional approval to charter any AI company valued at $1 billion or more; the Authority charters smaller ones; reapproval every 10 years; quarterly safety certifications signed by executives under penalty of perjury
- Safety: 'Quarantine Before Release', meaning high-risk systems need federal approval after testing in isolated government facilities; surprise red-teaming by government and certified nonprofits; mandated kill switch and emergency shutdown on order of federal inspectors; whistleblower bounties
- Capability bans: AI 'capable of evading human control, including by recursive self-improvement, self-modification, autonomous self-replication, and resistance to shutdown'; AI creating catastrophic CBRN or critical-infrastructure risk; AI replacing licensed professionals, skilled trades or human care providers
- Structure: AI companies barred from search and social media; a company may own AI infrastructure or AI systems, but not both; no ownership of non-AI businesses; ban on training on surveillance data and on using copyrighted works without authorization and compensation
- Governance and labor: independent Chief Safety Officer; board seats for a safety specialist, a union representative and a public-interest advocate; caps on incentive pay and no golden parachutes; voluntary union recognition as a condition of the 10-year charter
- Liability: strict liability plus mandatory insurance; personal civil and criminal liability for executives; 'corporate death penalty' (charter revocation and liquidation) for reckless or repeat offenders; private right of action
- Money: a Monopoly Data Tax on Big Tech firms that built AI on scraped or surveillance data, paid into an AI Public Fund that distributes 'AI Bonds' to all Americans
- Jayapal: 'If you want to be an AI company in this country, you have to obtain a public charter that has a broad set of terms and conditions to operate.' Endorsers include former FTC Chair Lina Khan, AFT, Public Citizen, Future of Life Institute, AI Now and the American Economic Liberties Project (Common Dreams; press release lists 24 groups)
What happened
Jayapal, a leading House progressive, released the most sweeping federal AI proposal of the 2026 debate. It treats AI companies like chartered banks or nuclear operators: a license to operate that is renewed every ten years, can be revoked, and comes with structural separation from Big Tech, pre-release federal testing, government kill switches and personal liability for executives.
Why it matters
It shows how far the Democratic left has moved since the September 2026 agent incidents and lab calls for pacing. The framework writes "rogue AI" bans (self-replication, resistance to shutdown) and nuclear-style containment directly into proposed law. It has no chance in the current Republican-controlled Congress, but it marks out one end of the debate before the November 2026 midterms. Watch for formal bill text and cosponsors.
Changelog
- 2026-10-04: created (sweep 2026-10-04; details from the framework PDF)
Related events
- Senators Hawley and Murphy announce the bipartisan AI Agent Accountability Act: criminal and civil CFAA liability for AI agent operators and developers over agent hacking ★★★
- Sanders and Casar introduce the Ban Artificial Superintelligence Act, with a pause on advanced AI and a new Department of AI ★★★
- Thune, Cruz and Klobuchar negotiate a Senate bill imposing a 'duty of care' on frontier AI developers and letting the government block unsafe model releases ★★★★
Sources (4)
- officialRep. Jayapal press release: legislative framework establishing a national charter system to rein in AI
- officialThe National AI Charter Act Framework (PDF)
- pressCommon Dreams: Under Jayapal bill, every AI company would need a public charter to operate in the US
- pressMLex: Federal charter system needed for AI safety, US Rep. Jayapal says
id: 2026-10-01-jayapal-national-ai-charter-act-framework · updated 2026-10-04 · open in the interactive timeline