The AI Whistleblower Protection Act is a bipartisan federal bill introduced in the U.S. Senate by Sen. Chuck Grassley (R-Iowa) on May 15, 2025. It would create explicit federal whistleblower protections for employees of AI developers and deployers, shielding disclosures of AI-related dangers and overriding restrictive non-disclosure and severance agreements. The bill has been read twice and referred to committee.
| Jurisdiction | United States (federal) |
| Bill ID | S.1792, 119th Congress (2025–2026); House companion H.R.3460 |
| Sponsor | Sen. Chuck Grassley (R-Iowa), Chair, Senate Judiciary Committee |
| Introduced | 2025-05-15 |
| Status | Introduced — read twice and referred to the Senate Committee on Health, Education, Labor, and Pensions |
Status and timeline
S.1792 was introduced in the 119th Congress on May 15, 2025, read twice, and referred to the Senate Committee on Health, Education, Labor, and Pensions. A House companion, H.R.3460, was introduced by Reps. Jay Obernolte (R-Calif.) and Ted Lieu (D-Calif.). As of the last status check, the congress.gov summary was still listed as "in progress"; the description of the bill's provisions reflects the bill's text and its sponsors' public account (Source: congress.gov).
Scope and provisions
The bill creates explicit federal whistleblower protections for people who develop and deploy AI. Its stated premise is that employees inside AI companies are often the first to see emerging dangers, but that restrictive severance agreements and non-disclosure agreements (NDAs) create a chilling effect that discourages them from disclosing what they know to the federal government, including Congress.
According to its sponsors, the bill merges existing AI and whistleblower-protection law to shield the communications of current and former AI employees who make protected disclosures, covering reports of significant dangers, security failures, AI security vulnerabilities, and breaches of law. It overrides restrictive NDAs and severance terms that would otherwise silence those disclosures. It also provides relief for AI whistleblowers who suffer retaliation, including reinstatement, back pay, and compensation for damages where applicable (Source: judiciary.senate.gov).
Existing federal whistleblower statutes are organized by sector or by specific fraud regimes, and AI-development risks such as capability dangers, security failures, and safety-evaluation suppression do not map cleanly onto any of them; the bill creates a sector-specific protected channel. The override of NDAs and severance terms is the mechanism by which the protected disclosures, currently subject to contractual suppression, would be made available. The measure functions as a complement to disclosure mandates such as California SB 53, whose enacted whistleblower provisions are the closest comparator: rather than requiring companies to publish information, it protects the employees who can reveal what companies do not. Radical Optionality: Governing Transformative AI Under Uncertainty (Winter & Bullock) document the Act as a capacity-building measure that strengthens the government's information channels about frontier AI without imposing heavy substantive regulation, consistent with the radical optionality argument.
Sponsors and support
The bill is bipartisan. Co-sponsors include Sens. Chris Coons (D-Del.), Marsha Blackburn (R-Tenn.), Amy Klobuchar (D-Minn.), Josh Hawley (R-Mo.), and Brian Schatz (D-Hawai'i). The House companion, H.R.3460, was introduced by Reps. Jay Obernolte (R-Calif.) and Ted Lieu (D-Calif.).
Endorsing organizations include the National Whistleblower Center, the Government Accountability Project, the Center for AI Policy, The Anti-Fraud Coalition, Encode, and Americans for Responsible Innovation, a coalition spanning whistleblower-advocacy and AI-safety-policy organizations (Source: judiciary.senate.gov).
Background
The bill grew out of Sen. Grassley's scrutiny of AI-company employment practices. Grassley previously sent a letter to OpenAI CEO Sam Altman raising concerns about allegedly illegally restrictive NDAs and the company's severance and non-disparagement agreements, the same chilling-effect concern the bill is designed to address. The legislation arrives amid a series of high-profile departures and public-letter campaigns by frontier-lab employees and former employees calling for stronger external oversight of AI development.
Relationships
- related: AI Whistleblowing — the concept page this bill anchors.
- related: California SB 53 — state frontier-safety law with its own whistleblower provisions; the closest enacted comparator.
- supports: Radical Optionality — a low-cost capacity-building measure of the kind the radical-optionality argument favors.
- related: Sen. Chuck Grassley — bill sponsor; Radical Optionality: Governing Transformative AI Under Uncertainty — documents the Act.
- related: OpenAI — the restrictive-NDA practices that motivated the bill.
Sources
- (Source: congress.gov) — Congress.gov, S.1792, 119th Congress: sponsor, introduction date, committee referral, status.
- (Source: judiciary.senate.gov) — U.S. Senate Committee on the Judiciary, "Grassley Introduces AI Whistleblower Protection Act" (May 15, 2025): bill purpose, co-sponsors, House companion, endorsements, background.
- Radical Optionality: Governing Transformative AI Under Uncertainty — Winter & Bullock document the Grassley AI Whistleblower Protection Act as an example capacity-building measure.