The letter of September 16, 2026 is a two-page letter to Senate Majority Leader John Thune, Senate Minority Leader Chuck Schumer and "Members of the U.S. Senate," written "as the U.S. Senate works to craft a framework for AI governance." It is signed by 20 organizations and five individuals, led by Americans for Responsible Innovation (ARI), and published on ARI's website under the title "Senate Coalition Letter - Redlines on AI Safety Legislation." It sets out three safeguards it calls "the minimum threshold that we would expect a framework for federal AI governance to meet": government-set, mandatory and enforceable safety standards for advanced AI; mandatory and independently conducted evaluation and auditing of advanced models; and a federal floor that preserves the states' authority to regulate further. It names no bill and no senator other than its addressees.
The three requirements
Government-set standards. The letter's central request is printed in bold: "Specifically, we strongly request that any AI regulatory framework empower the government, not AI companies, to set mandatory, binding, and enforceable safety standards for advanced AI." It states that "experience has proven time and again that tech companies cannot be allowed to set their own rules when it comes to the safety of their products," and, citing "our national security and the future of American communities," urges the leaders "to ensure that forthcoming AI governance legislation meets this minimum requirement."
Independent evaluation and auditing. "In addition to enabling mandatory, enforceable government-set standards for AI, a framework for AI governance should require substantive, transparent, and independent evaluations of advanced AI models and AI development." The letter states that "recent incidents of AI models escaping human oversight and even committing cybersecurity crimes highlight that we need more examination of models before they are deployed, both internally and externally." It describes recent voluntary transparency commitments by leading laboratories as "a step forward" that "are ultimately only voluntary and reach only as far as tech companies will allow them to," and concludes: "Auditing of AI models must be both mandatory and conducted independently." See AI Pre-Release Vetting.
A federal floor. "Finally, we strongly urge that Congress establish a strong floor with regard to AI regulations while preserving the critical rights of the states to establish additional regulations beyond those set by Congress. At a moment of rising public concern over quickly advancing AI risks, Congress should not eliminate strong state laws in exchange for weaker federal standards." See AI Federalism and Techno-Federalism.
The letter describes these three as "a necessary foundation for a federal AI framework on frontier AI risks," and names "whistleblower protections" and "additional child chatbot safety requirements" as other measures that "deserve Congress's attention" beyond its scope.
Stated grounds
The letter opens by stating that "this summer's wave of multi-agent AI incidents and multiple whistleblower accounts demand a strong response from the U.S. Senate," and that "it should now be clear that the most advanced AI systems present substantial catastrophic and plausibly existential risks to humanity." It acknowledges that "AI companies have made progress on voluntary agreements to provide visibility into frontier development," but argues that this transparency "will continue to fall short if it is not paired with binding federal standards."
Its account of the companies' conduct is that "leading AI companies have now broken the public trust time and again by releasing products that endanger users," that "AI companies have not been fully transparent about loss-of-control incidents with unreleased models," and that "neither the public nor lawmakers have any meaningful insight into what these models can do and whether or not companies are following best practices or even their own commitments to security and safety." It states that "while America's AI labs are global leaders in innovation, they cannot be trusted with setting the rules that protect public safety."
The letter does not identify the incidents, whistleblower accounts, voluntary agreements or products it refers to.
Signatories
The letter lists 20 organizations and five individual signatories, 25 in all.
Organizations, in the order listed: The Alliance for Secure AI; Americans for Responsible Innovation; the Center for AI Safety Action Fund; Common Cause; CommonDefense.us; Common Sense Media; Consumer Reports; Demand Progress Action; Design It For Us; Economic Security Project Action; Encode AI; FAR.AI; the Future of Life Institute; Humans in Control; the Institute for Family Studies; Issue One; Public Citizen; Public Knowledge; the Secure AI Project; and The Tech Oversight Project.
Individuals, as the letter identifies them: Alex Bores, New York State Assemblyman, District 73; Malo Bourgon, Machine Intelligence Research Institute; Daniel Kokotajlo, AI Futures Project; Girish Sastry, Guidelight; and Alexander Turner, AI researcher. The letter gives no further affiliation for Turner, and does not establish whether he is the former Google DeepMind researcher Alex Turner; nor does it establish whether Sastry's "Guidelight" is Guidelight AI Standards.
Seven of the organizations (ARI, the Alliance for Secure AI, Common Cause, Design It For Us, Public Citizen, Public Knowledge and The Tech Oversight Project) also signed ARI's letter of September 9, 2026 to President Trump, and Demand Progress signed that letter while Demand Progress Action signs this one (Coalition letter to President Trump on public release of the White House frontier-AI review framework (September 2026)).
Context
The letter was sent while a frontier-AI bill developed by Thune and Sen. Amy Klobuchar, with Commerce chair Ted Cruz also working on it, remained unreleased; coverage described it as resting on a "duty of care" that would expose developers to legal liability for failing to manage catastrophic risks (Source: rollcall.com; washingtonpost.com). On the day of the letter, Cruz said he would "love to mark it up this month, if we can get bipartisan agreement," Klobuchar said she was "continuing to negotiate" the text, and Thune said it "remains to be seen" whether it would get a committee vote (Source: semafor.com). See AI Liability and Coordinated slowdown proposals compared.
The same day Rep. Sam Liccardo's office released a letter from 107 House Democrats asking Speaker Mike Johnson to keep the House in session until Congress advances AI safeguards (Letter from 107 House Democrats to Speaker Mike Johnson on AI safeguards (September 2026)). ARI's own federal blueprint of August 10, 2026 places assurance with the government, with accredited independent verification organizations able to augment compliance work in domains the regulator certifies (Responsible Innovation at the Frontier (ARI, August 2026)).
Coverage and how it differs from the letter
The Washington Post reported the letter as an exclusive on September 16, 2026, describing "a coalition of AI researchers and 19 advocacy groups" in "a campaign to reject legislation being developed by" Thune and Klobuchar. Its account, of which only the summary and quoted excerpts were retrievable, reports that AI safety advocates want embedded evaluators in the frontier laboratories, and that some conservatives and pro-innovation analysts instead expect the duty of care to help guard against regulatory capture by the laboratories (Source: washingtonpost.com). Inside AI Policy, which identified ARI as the letter's lead and linked the PDF, described it as having "two dozen signatories" (Source: insideaipolicy.com).
| Point | Coverage | Letter text |
|---|---|---|
| Signatory count | "19 advocacy groups" and AI researchers (Washington Post); "two dozen signatories" (Inside AI Policy) | 20 organizations and 5 individuals |
| Object | A campaign "to reject legislation being developed by" Thune and Klobuchar (Washington Post) | Sets minimum requirements for "a framework for AI governance" the Senate is crafting; names no bill; does not mention Klobuchar or a duty of care |
| Evaluators | Advocates want evaluators embedded in the laboratories (Washington Post) | Asks for evaluations and auditing that are "mandatory and conducted independently"; does not specify embedded evaluators |
| Addressees | Senate leadership | Thune, Schumer and "Members of the U.S. Senate" |
The retrieved coverage does not explain the Post's count of 19. Embedded evaluators are the first step of Dario Amodei's pacing essay of September 12, 2026, which places them with third parties invited by the company rather than with the government (We Must Pace the Frontier (Amodei, September 2026)).
Provenance
Captured on 2026-09-18 from the PDF hosted on ARI's domain (application/pdf, 106,764 bytes, Last-Modified September 16, 2026 00:25:24 GMT; two pages with a text layer); both pages, including the full signature list, were transcribed from the text layer. Inside AI Policy (September 18) links the same PDF as the letter "led by Americans for Responsible Innovation," and the Washington Post's exclusive of September 16 quotes the letter's bolded request verbatim. Earlier searches on September 17 and 18 had found only the Washington Post's account and no hosted copy. The letter is distinct from ARI's 28-organization letter of September 9, 2026 to President Trump asking the White House to publish its frontier-AI review framework (Coalition letter to President Trump on public release of the White House frontier-AI review framework (September 2026)).
Relationships
- related: Americans for Responsible Innovation (ARI) — lead signatory and host
- related: John Thune, Chuck Schumer — addressees
- related: Amy Klobuchar, Ted Cruz — developing the Senate framework the letter addresses, which it does not name
- related: AI Pre-Release Vetting — mandatory, independently conducted evaluation and auditing
- related: AI Federalism, Techno-Federalism — the federal floor and preservation of state authority
- related: AI Liability — the duty-of-care framework that coverage reported the letter as rejecting
- related: Coordinated slowdown proposals compared — congressional responses to the September 2026 pacing debate
- related: Letter from 107 House Democrats to Speaker Mike Johnson on AI safeguards (September 2026) — the House letter released the same day
- related: Coalition letter to President Trump on public release of the White House frontier-AI review framework (September 2026) — ARI's earlier letter to the President, a separate document
- related: Responsible Innovation at the Frontier (ARI, August 2026) — ARI's federal blueprint
- related: The Alliance for Secure AI, Encode AI, Future of Life Institute (FLI), Center for AI Safety (CAIS), Public Citizen, Secure AI Project, Machine Intelligence Research Institute (MIRI), Alex Bores, Daniel Kokotajlo, AI Futures Project — signatories or their affiliations
Sources
- Raw source:
Raw Sources/Senate Coalition Letter - Redlines on AI Safety Legislation (ARI, 2026-09-16).md - Letter PDF: ari.us
- Coverage: washingtonpost.com; insideaipolicy.com