The JAWBONE Act (S. 4749) is a bipartisan federal bill introduced by Sen. Ted Cruz (R-TX) and Sen. Ron Wyden (D-OR) on June 11, 2026 that would bar federal agencies and employees from coercing broadcasters, online services, or AI services into changing content. The bill extends longstanding "jawboning" concerns — informal government pressure on private speech intermediaries — to AI services explicitly, placing AI model providers alongside broadcasters and online platforms as protected parties (Source: arstechnica.com).
Key provisions
As described in reporting on its introduction, the bill would:
- Prohibit coercion. Bar federal agencies and employees from coercing broadcasters, online services, or AI services into changing content.
- Create a private right of action. Allow affected parties to sue for compensatory damages regardless of whether the coercion succeeded in changing content.
- Require transparency logging. Require agencies to log their communications with AI and social media companies in a public portal.
(Source: arstechnica.com)
Committee markup
The Senate Commerce Committee took up the bill at a markup of several bills on September 17, 2026. Ranking member Maria Cantwell (D-WA), who said she supports its goals and cited the episode involving Federal Communications Commission chairman Brendan Carr, ABC and Jimmy Kimmel as a case in which "a speaker should have a real remedy when the government leans on a platform to silence them," argued that because there is "essentially, right now, no federal law governing AI," under the chairman's text "almost any government conversation with an AI company [about] safety could be called coercion." Her example was a Department of Energy employee who asks a company to stop its model from explaining how to enrich uranium: that request would be "content action with no existing federal law behind it," so the communication "could be illegal under this bill, and the employee could be sued." She offered a substitute amendment that would remove AI models from the bill, strike the public portal of government communications, limit liability to decision-making officials, and require a judge to find any consent judgment in the public interest. The amendment was not agreed to, and Cruz committed to work with her to address the concerns (Source: commerce.senate.gov). Inside AI Policy reported that Cruz committed to meet with committee colleagues concerned that the bill could chill the speech of officials trying to mitigate national security risks from misuse of AI systems, and possibly to adjust the bill as it advances; only the lede of that account was retrievable (Source: insideaipolicy.com).
The objection had been put to the committee two days earlier. In a letter dated September 15, 2026 to Cruz and Cantwell, Americans for Responsible Innovation president Brad Carson wrote that the bill's cause of action reaches any government communication that seeks to influence a content decision by an AI developer, social media company or broadcaster, exempting only lawful investigations and actions required or authorized under existing law, and that "because there are hardly any federal laws that directly regulate AI risk, most warnings a federal official might give an AI developer about a safety risk would fall outside these two exemptions." The letter welcomed an amendment Cruz had offered adding a rule of construction that the bill does not prevent a federal official from discussing national security risks posed by AI with an AI developer, but said concerns remained because "national security risk" is hard to define, citing the OpenAI agents' breach of Hugging Face as an incident that "may or may not meet a given definition of a national security risk." It concluded that an official facing possible civil liability for a fact-specific judgment would tend "to avoid the conversation entirely. If this happens, the American people will be less safe" (Source: commerce.senate.gov).
Support
The American Civil Liberties Union (ACLU), the Foundation for Individual Rights and Expression (FIRE), the Knight First Amendment Institute, and Americans for Tax Reform back the bill (Source: arstechnica.com).
Context
By naming "AI services" as a protected category, the bill situates AI model providers within the jawboning debate that has previously centered on social media content moderation. It runs alongside, and in some tension with, the contemporaneous federal push for government visibility into and access to frontier models under the June 2 AI cybersecurity executive order and NSPM-11, which encourage developers to give the government pre-release model access. Where those measures expand the channels between AI labs and federal agencies, the JAWBONE Act would constrain how agencies may use such channels to influence model outputs.
Relationships
- related: EO — Promoting Advanced AI Innovation and Security (Trump, signed June 2, 2026), AI and Surveillance, AI and the First Amendment
- supports: AI and the First Amendment — extends speech-intermediary protections to AI services
- related: Ted Cruz — lead sponsor and Commerce Committee chair; Maria Cantwell — ranking member whose substitute amendment would have removed AI models
- related: Americans for Responsible Innovation (ARI) — September 15, 2026 letter warning the bill could deter officials from raising AI safety risks with developers
Sources
- Ars Technica, June 11, 2026: "Ted Cruz and Ron Wyden try to fight censorship with bipartisan JAWBONE Act" — (Source: arstechnica.com)
- Senate Commerce Committee (minority), September 17, 2026: "Cantwell Receives Commitment to Refine JAWBONE Act at Markup of Legislation" — (Source: commerce.senate.gov)
- Americans for Responsible Innovation, letter to Chairman Cruz and Ranking Member Cantwell, September 15, 2026 — (Source: commerce.senate.gov)
- Inside AI Policy, September 21, 2026: "Cruz to look at finetuning JAWBONE Act to address AI-related national security risks" (lede only) — (Source: insideaipolicy.com)