Legal challenges to AI regulation and First Amendment claims

Legal Challenges to AI Regulation and First Amendment Claims

The United States has navigated a complex path in regulating artificial intelligence, balancing innovation with accountability. Federal agencies like DARPA, the National Science Foundation, and NASA spearheaded early research, while state-level engagement remained minimal during the 1950s–2010s. The Obama administration prioritized AI risk assessment, culminating in the 2016 Preparing for the Future of Artificial Intelligence report. By the 2010s, the National Strategic Research and Development Plan for Artificial Intelligence (2016) and Section 1051 of the 2018 National Defense Authorization Act established the National Security Commission on Artificial Intelligence (NSCAI). President Joe Biden’s 2022 AI Bill of Rights outlined protections for safe systems, data privacy, and human oversight.

As AI’s societal impact grew, states enacted targeted legislation. California’s SB1047 (2024) mandated transparency, data protection, and algorithmic fairness, while Illinois’s HB3773 (2025) prohibited AI-driven hiring decisions with discriminatory outcomes. Colorado’s AI Act (2025) aimed to reduce algorithm,ic discrimination in healthcare and finance. Meanwhile, President Donald J. Trump’s 2025 AI Action Plan sought to centralize federal authority, promoting open-source models and reducing state-level regulations.

Global and Industry Dynamics

U.S. regulatory efforts intersected with global initiatives. The European Union’s proposed AI Act (a tiered framework for AI classification) and China’s 2021 AI regulations, emphasizing information control and worker protections, highlighted divergent approaches. Domestically, the National Institute of Standards and Technology (NIST) released the AI Risk Management Framework (AI RMF 1.0) in 2023. Industry self-regulation, such as the Frontier Model Forum (2023), aimed to set safety standards for advanced models.

Public and Political Reactions

Public sentiment remained mixed, with a 2022 Pew Research Center survey showing 18% of Americans expressed greater concern than excitement about AI. State governments, such as New York City’s Bias Audit Law (2023), addressed algorithmic bias, while others like Texas’s Responsible AI Governance Act (2025) prohibited AI systems that incite violence or unlawful discrimination. Federal and state tensions persisted, with the National Conference of State Legislatures (NCSL) advocating intergovernmental collaboration.

Balancing Free Speech and AI Regulation

The intersection of AI regulation and First Amendment protections has become a focal point of legal debate, with states and federal authorities clashing over permissible content moderation and speech restrictions.

State-Level AI Laws and Free Speech Concerns

Several states have enacted AI-specific legislation, raising questions about their compatibility with First Amendment principles. California’s Transparency in Frontier AI Act (SB 53), effective January 1, 2026, mandates risk frameworks and safety reporting for AI systems with over 10²⁶ FLOPS, with penalties up to $1 million per violation. Critics argue such requirements could compel content moderation, potentially chilling speech under the guise of safety protocols. Texas’s Responsible AI Governance Act (HB 149) prohibits AI systems designed for restricted purposes, such as encouraging self-harm or generating child sexual abuse material (CSAM), with penalties ranging from $10,000 to $200,000. Legal scholars note these provisions may inadvertently restrict lawful speech, depending on enforcement mechanisms.

Illinois’s Human Rights Act amendment (HB 3773) classifies AI use in employment decisions without notice or discriminatory practices as civil rights violations, granting a private right of action. This law could conflict with employers’ First Amendment rights to use AI tools for hiring, though its scope remains under litigation. Colorado’s AI Act (SB 24-205), set to take effect June 30, 2026, requires impact assessments and algorithmic discrimination prevention, further complicating the balance between regulatory oversight and free expression.

Federal Preemption and Legal Challenges

President Trump’s December 11, 2025, executive order preempted state AI laws, specifically naming the Colorado AI Act as a target for federal oversight. The Department of Justice (DOJ) established an AI Litigation Task Force in January 2026 to challenge state regulations, citing the executive order’s intent to reduce regulatory burdens on start-ups. This federal push has sparked lawsuits, including Amazon v. Perplexity, where Amazon alleges Perplexity violated federal and California computer fraud statutes by using its AI agent, Comet, to interact with Amazon.com without disclosing its identity. Amazon seeks a preliminary injunction, while Perplexity claims the suit aims to stifle competition.

The TAKE IT DOWN Act (signed May 19, 2025) mandates notice-and-removal processes for non-consensual intimate imagery, including AI-generated deepfakes, by May 19, 2026. While this law targets specific harms, its implementation could intersect with First Amendment protections for expressive content, particularly if broadened to include political or artistic speech.

International and Legal Frameworks

The European Union’s Digital Omnibus proposal (November 19, 2025) restructured compliance timelines for high-risk AI systems, with penalties for non-compliance extending to December 2, 2027. The GDPR amendment adds Article 88c, permitting legitimate interest as a legal basis for AI data processing with safeguards, though its application to free speech remains unclear.

In the U.S., the FTC vacated its 2024 consent order against Rytr LLC, citing the Trump Administration’s AI Action Plan, which prioritizes reducing regulatory burdens. This decision underscores the shifting legal landscape and its impact on how AI-related claims are adjudicated.

Legal Hurdles in Regulating AI Technologies

The legal landscape for regulating AI technologies in 2026 is marked by fragmented frameworks, evolving liability standards, and cross-jurisdictional conflicts, creating significant hurdles for compliance and innovation.

EU AI Act Implementation and U.S. State-Level Fragmentation

The European Union’s AI Act, which entered phased implementation in 2025, imposes obligations on general-purpose AI (GPAI) models, requiring providers to publish training data summaries and restrict untargeted facial scraping. As of 2026, downstream users must ensure their systems avoid prohibited categories, such as biometric surveillance, to prevent supply chain disruptions. Meanwhile, the U.S. lacks federal AI legislation, leaving states to drive regulation. California, Texas, and Colorado have enacted stringent laws: the Colorado AI Act, effective June 2026, mandates reasonable care impact assessments, while Texas’s TRAIGA law bans AI systems designed to incite self-harm or produce deepfakes, requiring disclosures for government and healthcare AI use.

Intellectual Property and Liability Challenges

Litigation over AI’s use of copyrighted data, such as NYT v. OpenAI and Getty v. Stability AI, is nearing decisive rulings. Courts are assessing whether training on copyrighted material constitutes fair use, with adverse outcomes potentially forcing licensing regimes or deployment limits. Additionally, AI agents—capable of executing code, signing contracts, and booking transactions—pose novel liability questions. Courts have not yet resolved whether users or developers bear responsibility for autonomous errors, prompting organizations to revise vendor contracts to include indemnification clauses for AI-generated financial losses.

Data Privacy and the “Right to Unlearn”

Privacy regulators are scrutinizing the permanence of large language models, questioning whether deleting user data from databases suffices if the data remains embedded in model weights. Organizations are advised to update privacy policies to clarify technical limitations of deletion requests, as the legal status of data erasure in trained models remains unresolved.

Outbound Investment Restrictions

U.S. Treasury rules, effective January 2025, restrict outbound investments in foreign entities, particularly in China, for AI developments with military or surveillance applications. Venture capital firms must now vet portfolio companies for exposure to restricted technologies, adding complexity to global AI investment strategies.

Federal-State Conflicts and Global Variance

President Trump’s 2025 executive order directs the attorney general to challenge state laws conflicting with a “minimally burdensome national policy framework,” creating regulatory uncertainty. Meanwhile, the UK enforces stricter transparency in AI decision-making for sectors like credit and hiring, while Canada’s Artificial Intelligence and Data Act (AIDA) imposes risk mitigation and transparency obligations on high-impact systems. China’s regulations emphasize algorithm governance and data security, with intensified enforcement expected in 2026.

Future Implications and Policy Recommendations

The legal and regulatory landscape surrounding AI in 2026 reflects a complex interplay between federal preemption, state innovation, and constitutional protections, with significant implications for technology development, free speech, and consumer rights.

Legislative Landscape: Federal vs. State Tensions

U.S. federal government, under President Donald Trump, has prioritized a unified national approach to AI regulation through the National AI Legislative Framework (March 20, 2026). This nonbinding roadmap, aligned with Executive Order 14365, seeks to preempt state laws deemed “onerous” by the Commerce Department, which delayed submitting its evaluation of state regulations until March 11, 2026. Key priorities include child safety, intellectual property (IP) protection, and free speech safeguards, while opposing vague standards or fragmented oversight. For example, the framework explicitly names the Colorado AI Act (SB 24-205) as a target for federal preemption, despite its provisions for impact assessments and algorithmic transparency.

State legislatures, however, have advanced robust AI laws. California’s Transparency in Frontier AI Act (SB 53) (2025) mandates risk frameworks and whistleblower protections for large AI models, with penalties up to $1 million per violation. Texas’s Responsible AI Governance Act (HB 149) prohibits AI systems designed for restricted purposes, such as encouraging self-harm or generating child sexual abuse material (CSAM), with penalties ranging from $10,000 to $200,000. Meanwhile, New York’s RAISE Act (2025) enacts AI safety legislation despite federal preemption efforts, highlighting the persistence of state-level innovation.

Regulatory Actions: FTC and DOJ Enforcement

The Federal Trade Commission (FTC) has intensified enforcement against deceptive AI practices, exemplified by its $18 million settlement with Air AI (March 24, 2026). The FTC alleged that Air AI falsely claimed users would achieve substantial earnings and misrepresented refund guarantees, violating the Telemarketing Sales Rule (TSR) and Business Opportunity Rule. The order bans Air AI from marketing business opportunities and making unsubstantiated earnings claims, underscoring the agency’s focus on consumer protection.

The Department of Justice (DOJ) has also moved to challenge state AI laws through its newly established Artificial Intelligence Litigation Task Force (January 9, 2026). Directed by Trump’s executive order, the task force aims to reduce regulatory burdens on start-ups and emerging tech firms, potentially leading to litigation over state laws like California’s SB 53 or Texas’s HB 149.

First Amendment Considerations: Free Speech vs. Content Moderation

AI regulation intersects with First Amendment debates, particularly in content moderation and IP protections. The FTC’s enforcement actions, such as the Air AI case, raise questions about whether restrictions on false claims infringe on corporate speech. Conversely, the TRUMP AMERICA AI Act (proposed by Sen. Marsha Blackburn) includes provisions to safeguard free speech while addressing harmful AI use, such as banning AI-washing (misleading claims about product efficacy).

The EU AI Act amendments (2026) further complicate the global context, as the GDPR’s revised Article 88c permits incidental processing of special category data in AI training, balancing innovation with privacy safeguards. However, U.S. critics argue that the EU’s approach risks stifling AI development through overly restrictive data rules.

Policy Recommendations: Balancing Innovation and Accountability

To navigate these tensions, policymakers should consider the following:

1. Clarify Federal Preemption Boundaries: The Commerce Department’s delayed evaluation of state laws (due March 11, 2026) highlights the need for transparent criteria to determine which state regulations conflict with federal priorities. A hybrid model, allowing limited state innovation while preventing duplicative oversight, could mitigate compliance costs.

2. Strengthen AI Transparency Standards: The Colorado AI Act’s requirement for impact assessments and consumer disclosures offers a template for federal legislation. Mandating algorithmic audits and bias detection, as proposed in the EU’s amendments, could address equity concerns without stifling innovation.

3. Enhance FTC Enforcement Tools: The FTC’s success in the Air AI case demonstrates the agency’s capacity to address deceptive practices. Expanding its authority to penalize AI-generated misinformation or deepfakes, while preserving free speech, would align with the National AI Legislative Framework’s emphasis on consumer protection.

4. Promote Bipartisan Collaboration: The GUARDRAILS Act (proposed by Democrats) and the TRUMP AMERICA AI Act reflect divergent approaches to regulation. A compromise, such as establishing a federal AI oversight body with input from states and industry stakeholders, could bridge partisan divides.

5. Support Workforce and IP Protections: The Framework’s focus on IP preservation and workforce readiness is critical. Voluntary licensing mechanisms for AI-generated content, as outlined in the Framework, could protect creators while enabling innovation, avoiding the pitfalls of overly prescriptive rules.

AI in Judicial Systems

The New York Courts’ AI Committee Report (2025 Annual Report) outlines interim policies for AI use in courts, emphasizing transparency, bias mitigation, and procedural fairness. These guidelines aim to ensure AI tools support judicial decision-making without compromising due process or equitable access to justice.