AI Governance Takes Shape: Breaking Down Washington’s Latest AI Frameworks

In a matter of days, Washington saw a flurry of competing artificial intelligence (AI) policy frameworks, signaling that the push for comprehensive federal AI legislation has entered a new consequential phase. Most notably, the White House released its National AI Policy Framework. This nonbinding yet highly influential legislative roadmap calls for targeted action on children’s safety, copyright, community protections and American AI competitiveness, while advocating for federal preemption of most state AI laws. Sen. Marsha Blackburn (R-TN) also unveiled the discussion draft for her national AI framework, the TRUMP AMERICA AI Act, staking out a broader and more prescriptive position than recommended by the White House.

Below is a detailed overview and comparison of key proposals, along with an assessment of next steps as congressional leadership and key committees begin developing their responses to the administration’s recommendations.

White House National Policy Framework on AI

The White House’s National Policy Framework on AI takes a middle-ground approach, addressing key concerns around children, copyright, conservatives and communities, the “four C’s,” while advocating for a light-touch regulatory posture and federal preemption of most state AI laws. The document is a nonbinding legislative recommendation that will likely carry significant influence as Congress moves to broadly address AI regulation. It does not prescribe a specific legislative vehicle, but key elements are expected to inform bills already circulating among Republicans.

A.Protecting Children and Empowering Parents

The framework urges Congress to equip parents with the tools to manage their children’s online privacy, screen time, content and accounts. It calls for age-assurance requirements for certain AI platforms and mandates safety features to reduce risks of sexual exploitation and self-harm. Importantly, it urges Congress to preserve states’ authority to enact their own child protection laws, including prohibitions on child sexual abuse material (CSAM).

B.Safeguarding and Strengthening American Communities

The framework addresses four key areas under this pillar:

• Energy Costs. Calls for codifying the Trump administration’s Ratepayer Protection Pledge to shield consumers from electricity cost increases driven by data center expansion and for streamlining federal permitting for AI infrastructure.
• Scams. Urges Congress to strengthen efforts combating AI-enabled impersonation and fraud, particularly targeting vulnerable populations.
• National Security. Calls for ensuring national security agencies have the technical capacity to assess frontier AI models and associated risks.
• Small Businesses. Urges Congress to provide resources to small businesses to support AI adoption and deployment.

C.Respecting Intellectual Property Rights and Supporting Creators

The administration takes the position that training AI on copyrighted material does not violate existing copyright law, while leaving final resolution to the courts. It calls on Congress to consider a licensing framework for collective rights systems and to establish federal protections against unauthorized distribution or commercial use of AI-generated replicas of a person’s voice, likeness or other attributes, with clear exceptions for parody, satire, news reporting and other First Amendment-protected expression. Congress is urged to monitor fair use developments and fill any gaps if needed, while refraining from near-term action that could interfere with ongoing judicial resolution.

D.Enabling Innovation and Ensuring American AI Dominance

The framework calls for regulatory sandboxes to advance U.S. AI leadership, federal dataset accessibility for training purposes and explicitly directs Congress not to create a new federal rulemaking body to regulate AI.

E.Educating Americans and Developing an AI-Ready Workforce

The framework urges Congress to integrate AI into education and workforce training through nonregulatory means, and to expand research into task-level workforce shifts driven by AI.

F.Establishing a Federal Policy Framework, Preempting Cumbersome State AI Laws

The framework directs Congress to preempt overly burdensome state AI laws while preserving traditional state police powers, zoning authority and states’ ability to govern their own use of the technology. It cautions that preemption must cover areas best suited to a national standard, including rules governing AI development, Americans’ use of AI and liability frameworks that would penalize developers for third-party misuse.

TRUMP AMERICA AI Act Discussion Draft – Sen. Marsha Blackburn (R-TN)

Sen. Marsha Blackburn’s (R-TN) bill proposes a comprehensive federal framework governing AI, online platforms and digital content with the stated goal of promoting AI innovation while safeguarding children, workers, creators, consumers and national security. The legislation is intentionally expansive, combining technology policy, labor reporting, platform accountability, child safety, intellectual property (IP) and national security oversight into a single statutory structure.

The bill applies broadly to companies and entities engaged in the development, deployment or operation of AI technologies, particularly AI chatbots, advanced AI systems and algorithmically driven online platforms. It relies on a multiagency enforcement and coordination model, with responsibilities allocated across the Federal Trade Commission (FTC), Department of Justice (DOJ), Department of Energy (DOE), National Institute of Standards and Technology (NIST), Department of Labor (DOL), Department of Commerce (DOC) and the National Science Foundation (NSF). While the bill establishes significant new federal standards, it largely preserves state authority through savings clauses, preempting state law only where direct conflicts arise.

This bill is best understood as a negotiating tool with the Trump administration and as congressional negotiations accelerate. The bill is strategically broad and designed to preserve Sen. Blackburn’s status as a key negotiator for any legislative product.

A.AI Chatbots and Duty of Care (Title I)

A central component of the bill is the creation of a federal duty of care for AI chatbot developers. Developers would be required to exercise reasonable care to prevent and mitigate foreseeable harms caused by their systems, with liability assessed under a negligence-style standard. The bill empowers the FTC to issue rules defining minimum safeguards and to enforce violations as unfair or deceptive practices. State attorneys general would also be empowered to bring enforcement actions, raising questions about the extent to which AI developers could face overlapping federal and state scrutiny.

B.AI and Workforce Impacts (Title II)

The bill addresses the labor impacts of AI by mandating regular disclosure of AI-related job effects. Covered entities, including publicly traded companies, federal agencies and certain private firms, would be required to report data on layoffs, hiring, retraining and unfilled positions attributable to AI adoption. The bill tasks DOL with collecting, analyzing and publicly releasing this information.

C.Repeal of Section 230 (Title III)

The bill sunsets and repeals Section 230 of the Communications Decency Acy after a two-year delay. The repeal eliminates a long-standing immunity for online platforms with respect to third-party content, exposing companies to substantially heightened litigation and compliance risk.

D.Protecting Children Online (Title IV)

Child protection is a major focus of the legislation. The bill establishes a statutory duty of care for platforms likely to be used by minors. Platforms must implement default safety protections, limit personalized recommendation systems that encourage compulsive use, provide robust parental controls, and maintain reporting and response mechanisms for harmful content. Covered platforms would be subject to transparency audits and enforcement by the FTC and state attorneys general.

E.AI Companions and Criminal Provisions (Title V – GUARD Act)

The bill further targets AI systems that simulate emotional or interpersonal relationships through the GUARD Act, which restricts so-called AI companions. Minors would be prohibited from accessing these systems, and covered entities would be required to deploy meaningful age-verification processes to distinguish adults from minors.

F.Risk-Based Framework for Advanced AI (Title VI)

The bill establishes a risk-based regulatory framework for advanced AI systems, defined by computational training thresholds. Developers of such systems would be required to participate in a DOE-led evaluation program involving pre-deployment testing, reporting and risk assessment. Noncompliance would trigger severe civil penalties.

G.AI Liability and Consumer Protection (Title VII – AI LEAD Act)

The bill establishes a federal AI liability regime under the AI LEAD Act. It clarifies standards for developer and deployer liability, introduces strict liability for defective or unreasonably dangerous AI systems, and prohibits contractual provisions that waive or unduly limit liability. A federal cause of action would allow such suits by individuals, classes, states and the federal government. Foreign AI developers would also be required to register U.S. agents, creating accountability for overseas firms accessing the U.S. market.

H.Bias, Transparency, and Censorship Concerns (Title VIII)

Concerns over bias and censorship are addressed through mandatory audits of high-risk AI systems, with a particular emphasis on political viewpoint and affiliation discrimination. Covered entities would be required to conduct annual independent audits and provide ethics training for relevant personnel.

I.AI Innovation, Standards, and Research (Titles IX–XI)

Alongside the regulatory components, the bill invests heavily in AI innovation and infrastructure. It expands federal testbeds, research programs and standards development led by NIST, promotes international cooperation with allied nations, and restricts access for foreign adversaries.

J.Individual Voice and Visual Likeness Protections (Title XII – NO FAKES Act)

Title XII of the bill includes the NO FAKES Act, which establishes a new federal intellectual‑property‑style right that protects an individual’s voice and visual likeness against unauthorized digital replication. This is specifically geared toward highly realistic, computer‑generated likenesses while carving out exceptions for legitimate uses such as authorized remastering or editing. The bill creates civil liability for publicly distributing, displaying or offering products or services that use an unauthorized digital replica, as well as for distributing tools primarily designed to create unauthorized replicas.

K.Rightsholder Subpoena Power (Title XIII – TRAIN Act)

Title 13 creates a subpoena mechanism that authorizes courts to compel the production of documents “sufficient to identify with certainty the copyrighted works … that were used by the developer to train the generative artificial intelligence model” when asked to do so by rightsholders. Courts are directed to grant the subpoena request so long as the subpoena is in proper form and includes a properly executed declaration stating that the rightsholder has a subjective good faith belief that the developer used some or all of at least one or more copyrighted works. Developers that fail to comply absorb a presumption that they made copies of copyrighted works. Rightsholders are prohibited from sharing the records they obtain, but do not appear to be prohibited from sharing information describing those records.

Practically speaking, this would work as a pre-discovery tool for rightsholders that suspect their copyrighted works were being used by developers to train AI models.

L.Content Provenance Information Requirements (Title XIV)

The bill’s content provenance requirements can be split into three categories.

First, the bill directs federal authorities to facilitate the development of technical standards for provenance information and detection of synthetic or synthetically‑modified content. The bill assigns NIST responsibility for research, development and public education on provenance systems, synthetic content detection, and methods to authenticate or label digital content.

Second, the bill requires entities developing synthetic (or synthetically-modified) content to include content provenance information that indicates that the content is synthetic. The bill prohibits the removal of the content provenance information.

Third, the bill prohibits the nonconsensual training of an AI system with copyrighted material, so long as the material has content provenance information attached. The covered content is specifically the material described in 17 U.S.C. § 102. Notably, these protections only apply when content provenance information is attached.

This section allows for enforcement by the FTC (violations are considered unfair or deceptive acts or practices), state attorneys general and through private actions.

M.Fair Use (Title XV)

The bill clarifies that the “unauthorized reproduction, copying, or computational processing of copyrighted works” for virtually any AI purpose (training, fine-tuning, developing, or creating) is not considered fair use under section 107. Similarly, AI models created “through inference, distillation, or similar processes” are presumed to incorporate copyrighted material unless the developer can establish by clear and convincing evidence that the AI was developed using authorized materials or that the training information included no copyrighted expression. The bill clarifies that any generated content that reproduces or services from copyrighted works constitutes infringement.

Regarding derivative works, the bill clarifies that derivative works created with copyrighted expression included in the training data are deemed infringing works. Those works are not eligible for copyright protection.

Comparing the Frameworks

The White House’s National AI Policy Framework stakes out a middle ground between Sen. Blackburn’s sweeping proposal and narrower approaches, advocating for a light-touch regulatory posture, federal preemption of most state AI laws and targeted action across children’s safety, copyright, community protections and American competitiveness.

---

THIS DOCUMENT IS INTENDED TO PROVIDE YOU WITH GENERAL INFORMATION REGARDING AI POLICY FRAMEWORKS. THE CONTENTS OF THIS DOCUMENT ARE NOT INTENDED TO PROVIDE SPECIFIC LEGAL ADVICE. IF YOU HAVE ANY QUESTIONS ABOUT THE CONTENTS OF THIS DOCUMENT OR IF YOU NEED LEGAL ADVICE AS TO AN ISSUE, PLEASE CONTACT THE ATTORNEYS LISTED OR YOUR REGULAR BROWNSTEIN HYATT FARBER SCHRECK, LLP ATTORNEY. THIS COMMUNICATION MAY BE CONSIDERED ADVERTISING IN SOME JURISDICTIONS.