TRUMP AMERICA AI Act (Sen. Blackburn AI policy framework discussion draft)

Overview

The Take Responsibility for Unlocking the Machine Potential of America’s AI Act, commonly referred to as the TRUMP AMERICA AI Act (S. 5765), was introduced in the United States Senate in late 2024 by Senator Ted Cruz and several co-sponsors. This legislative initiative represents a significant pivot in the United States' approach to artificial intelligence governance, moving away from the comprehensive regulatory and safety-first framework established by the previous administration. The primary objective of the Act is to repeal Executive Order 14110, titled 'Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence,' which the bill's proponents argue has stifled innovation through burdensome reporting requirements and centralized bureaucratic oversight. By removing these perceived barriers, the Act seeks to restore a 'light-touch' regulatory environment that prioritizes rapid technological development and private-sector autonomy. The Act is framed as a necessary response to the global technological landscape, where the speed of development is seen as the primary factor in maintaining national security and economic competitiveness. Proponents argue that the previous administration's focus on 'safety' and 'equity' served as a 'backdoor' for government control over the private sector, whereas this Act seeks to return that control to the market. The legislation is not merely a repeal but a proactive statement of American intent to lead the world in AI development by providing the most favorable legal and economic environment for developers, researchers, and investors alike.

Definitions

The TRUMP AMERICA AI Act provides a specific set of statutory definitions designed to clarify the scope of federal authority and limit the potential for regulatory creep. Central to the Act is the definition of 'Artificial Intelligence,' which it aligns with existing statutory language but interprets through a lens of technical functionality rather than speculative risk. Unlike previous frameworks that might include broad socio-technical impacts in their definitions, S. 5765 focuses on the computational processes—such as machine learning, neural networks, and automated reasoning—that allow systems to perform tasks typically requiring human intelligence. This technical focus is intended to prevent the government from regulating traditional software or statistical tools under the guise of AI oversight. Another critical definition introduced or refined by the Act is that of 'Frontier Models.' The Act defines these as high-capability AI models that exceed specific computational thresholds or demonstrate emergent properties that could significantly impact national security or economic stability. However, the Act is careful to distinguish between the inherent capabilities of a model and its intended use, ensuring that the mere existence of a powerful model does not trigger immediate restrictive oversight. Additionally, the Act defines 'Adversarial AI' in the context of foreign threats, specifically identifying technologies developed or deployed by 'countries of concern' that are intended to undermine United States infrastructure, democratic processes, or military superiority. By clearly defining these terms, the Act establishes a legal boundary that protects domestic innovators while focusing federal resources on genuine external threats.

Governance and Institutional Framework

The governance structure proposed by the TRUMP AMERICA AI Act is characterized by a decentralization of authority and a reduction in the power of centralized 'AI Offices' within the executive branch. A key provision of the Act is the formal dissolution of the AI Safety Institute (AISI) as envisioned under the Biden administration, or at least a radical restructuring of its mandate. The Act shifts the focus of the National Institute of Standards and Technology (NIST) away from creating 'safety' benchmarks that might be used as regulatory requirements, and toward the development of voluntary technical standards that facilitate interoperability and performance measurement. This shift ensures that NIST remains a technical body rather than a de facto regulator. Under the Act, the Department of Commerce is designated as the primary federal agency responsible for coordinating AI policy, with a specific mandate to promote American competitiveness. The Act establishes an 'Interagency AI Coordination Council' composed of representatives from the Departments of Defense, State, and Energy, as well as the Office of the Director of National Intelligence. Unlike previous councils focused on ethics and safety, this body is tasked with identifying and removing regulatory bottlenecks that hinder the deployment of AI in critical sectors. Furthermore, the Act mandates that any new significant regulatory action related to AI by any federal agency must be submitted to Congress for approval, effectively stripping agencies of the power to create binding AI rules through the administrative rulemaking process without direct legislative consent.

Key Focus Areas

The TRUMP AMERICA AI Act focuses on three primary pillars: economic deregulation, national security supremacy, and the protection of free expression. In terms of deregulation, the Act prohibits federal agencies from requiring AI developers to submit their models for pre-deployment review or 'red-teaming' unless there is a specific, evidence-based threat to national security. This 'permissionless innovation' model is intended to allow American companies to iterate faster than international competitors. The Act also targets the use of federal procurement as a tool for social engineering, prohibiting the government from requiring AI vendors to adhere to 'woke' or 'biased' algorithmic standards that the bill's authors claim lead to the suppression of lawful speech or the promotion of specific political ideologies. On the national security front, the Act prioritizes the integration of AI into the United States military and intelligence apparatus. It authorizes increased funding for the 'Project Maven' style initiatives and streamlines the acquisition process for AI-driven defense technologies. The Act also emphasizes the protection of the AI supply chain, including semiconductor manufacturing and cloud infrastructure, from foreign interference. By treating AI as a 'dual-use' technology of supreme importance, the Act seeks to create a 'fortress' around American AI intellectual property while simultaneously encouraging its rapid deployment in the field. This focus extends to the international stage, where the Act directs the State Department to form 'AI Freedom Alliances' with like-minded nations to counter the digital authoritarianism of China and Russia.

Implementation Framework

The implementation of the TRUMP AMERICA AI Act relies on a series of 'deregulatory mandates' issued to federal agencies. Within 180 days of enactment, every major federal agency is required to conduct a comprehensive review of its existing rules and guidance to identify any provisions that unnecessarily impede the development or adoption of AI. Agencies must then submit a plan to the Department of Commerce and Congress to repeal or modify these rules. This 'bottom-up' approach to implementation is designed to ensure that the pro-innovation spirit of the Act permeates all levels of the federal bureaucracy, from the Federal Trade Commission to the Department of Transportation. To support the technical implementation of the Act, the legislation authorizes the creation of 'AI Innovation Zones.' These are physical or virtual environments where companies can test advanced AI systems with relative immunity from certain federal regulations, provided they maintain basic transparency with the Department of Commerce. This framework is intended to act as a more robust and less restrictive version of 'regulatory sandboxes' found in other jurisdictions. Additionally, the Act mandates the creation of a 'National AI Research Resource' (NAIRR) that is specifically optimized for high-performance computing and large-scale data sets, ensuring that academic researchers and small startups have the tools necessary to compete with large technology firms without being subject to the same reporting burdens as those firms.

Monitoring and Evaluation

Monitoring under the TRUMP AMERICA AI Act is focused on assessing the United States' relative position in the global AI race rather than monitoring the behavior of individual domestic companies. The Act requires the Secretary of Commerce to submit an annual 'State of American AI Leadership' report to Congress. This report must include metrics on private sector investment, the number of AI patents filed by U.S. entities versus foreign entities, and an assessment of the computational capacity available within the United States. If the report finds that the U.S. is losing ground in any critical AI sub-field, the Act triggers a requirement for the Interagency AI Coordination Council to propose new incentives or further deregulatory measures. Evaluation of the Act’s effectiveness is also tied to the impact of AI on the American workforce and economy. However, unlike other regulations that focus on the risks of automation, the TRUMP AMERICA AI Act evaluates success based on the creation of high-value jobs and the increase in national productivity. The Act establishes a 'Workforce Transition Task Force' that monitors the shift in labor markets but focuses its recommendations on retraining and education in AI-related fields rather than imposing restrictions on automation. This monitoring framework ensures that the government remains informed about the technological landscape while maintaining a hands-off approach to the actual development and deployment of the technology.

Penalties, Liability, and Appeals

The TRUMP AMERICA AI Act takes a unique approach to liability and penalties, primarily by seeking to limit the legal exposure of AI developers. The Act includes provisions that shield AI companies from certain types of civil liability if their models are used by third parties to commit unlawful acts, provided the company did not intentionally design the model for such purposes. This 'Safe Harbor' provision is intended to prevent a wave of litigation that could bankrupt startups or discourage the release of open-source models. The Act also clarifies that AI-generated content is not subject to the same liability standards as human-authored content in specific contexts, further protecting the platforms that host these models. In terms of penalties, the Act focuses on government accountability rather than private sector punishment. If a federal agency is found to be in violation of the Act’s deregulatory mandates—for example, by imposing unauthorized reporting requirements—the Act provides a streamlined 'Administrative Appeal' process for affected companies. Companies can petition the Department of Commerce or the Office of Management and Budget (OMB) to strike down the offending agency action. If the appeal is successful, the agency may face budget reallocations or other administrative sanctions. This reversal of the traditional penalty structure is a core feature of the Act’s 'America First' and 'Innovation First' philosophy, ensuring that the burden of compliance remains on the regulators rather than the regulated.

Relationship to Other Instruments

The TRUMP AMERICA AI Act is explicitly designed to override and replace the regulatory framework established by Executive Order 14110. Upon enactment, all directives, reporting requirements, and institutional bodies created solely by EO 14110 are nullified. This creates a clear legal break from the 'safety and equity' focus of the previous administration. The Act also seeks to preempt state-level AI regulations, such as California’s SB 1047 or similar measures, by declaring that AI systems engaged in interstate commerce are subject only to federal standards. This preemption is a critical component of the Act, as it prevents a 'patchwork' of state laws from creating a de facto national regulatory regime that is more restrictive than the federal one. Furthermore, the Act clarifies its relationship with existing civil rights and privacy laws. While it does not repeal the Civil Rights Act or the Privacy Act, it stipulates that these laws cannot be 'reinterpreted' by agencies to create new AI-specific obligations without explicit congressional authorization. For example, an agency cannot claim that an AI model is 'discriminatory' simply because its output does not meet a specific demographic quota, unless there is evidence of intentional disparate treatment. This ensures that the TRUMP AMERICA AI Act remains the primary lens through which all other federal laws are applied to the AI sector, maintaining a consistent and predictable legal environment for developers and users alike.

International Alignment

The international strategy of the TRUMP AMERICA AI Act is one of 'Strategic Competition' rather than 'Global Harmonization.' The Act rejects the idea of a global AI treaty or a centralized international regulatory body, such as those proposed by the United Nations or the European Union. Instead, it directs the United States to lead a coalition of democratic nations in developing technical standards that promote 'AI Freedom.' This involves ensuring that AI remains a tool for economic growth and individual liberty, in contrast to the 'surveillance AI' models favored by adversarial regimes. The Act explicitly discourages the United States from adopting the 'precautionary principle' prevalent in the EU AI Act, which it views as a recipe for technological stagnation. To achieve this alignment, the Act authorizes the Department of State to engage in bilateral and multilateral agreements that focus on data sharing, joint R&D, and the removal of trade barriers for AI technologies. These agreements are intended to create a 'Trusted AI Zone' among allies, where technology can flow freely without the risk of intellectual property theft or regulatory interference from non-aligned blocks. The Act also includes 'Reciprocity Clauses,' which state that the United States will only provide certain AI-related benefits or market access to countries that do not impose discriminatory regulations on American AI firms. This 'tit-for-tat' approach to international AI policy is designed to protect American interests and ensure that the global AI ecosystem remains favorable to the U.S. model of innovation.

Implementation Timeline

Compliance Checklist

Sources and References