American Federalism: April in Review
- Noah Farnsworth
- May 3
- 18 min read
Updated: May 5
Primary author: Noah Farnsworth
In This Edition
State AI and data privacy legislation and what it means for a national framework.
Recent Supreme Court decisions and their federalism implications.
Federalism Flashpoints
Thank you to everyone who has been following as we continue to reach a wider audience in the (never boring) federalism world. Each month we highlight a selection of the most important questions and controversies relating to states, federalism, and the precarious federal-state balance. Our focus will continue to be on monitoring federal and state activities that have implications for the principles of federalism.
April in Review
Are states really laboratories of experimentation? That ideal has come under scrutiny in recent years. But it also remains a vital and core principle of the American federal system in the age of AI. So...yes, no, and it depends...
This newsletter will highlight two important federalism controversies in April. We start by exploring recent AI and data privacy legislation across the nation. As the federal government appears to have backed away, temporarily, from their AI regulatory project, states have stepped into the void and pushed forward with hundreds of laws and experimental regulations that they view as crucial for AI safety, regulation, and innovation for the future.
We will then turn to three recent Supreme Court cases dealing with birthright citizenship, conversion therapy, and election law.
Introduction
On December 11, 2025, President Trump signed Executive Order 14365, "Ensuring a National Policy Framework for Artificial Intelligence." The order's purpose was threefold.
First, it sought to encourage the innovation and excellence of American AI companies, with the larger goal of promoting national and economics security "dominance."
To accomplish this first goal, the order revokes the Biden era attempt to "paralyze the industry." The EO noted that companies should be able to innovate "without cumbersome regulation" from the states. The order claimed that the states had created an unnecessary "patchwork" of different regulation and that they were"responsible for requiring entities to embed ideological bias within models." The EO also charged the states with sometimes "impermissibly regulating" beyond State borders, impinging on interstate commerce.
The order set a national framework that would forbid state laws that conflicted with national policy. This preemptive move would "ensure that children are protected, censorship is prevented, copyrights are respected, and communities are safeguarded."
The order came at an opportune time. Just months prior, Congress removed an AI provision from the "Big Beautiful Bill" that would have functioned quite similar to EO 14365.
The response to the EO, not surprisingly, has been energetic. Public officials, state attorneys general, and policy groups warned that Executive Order 14365 “risks establishing de facto national standards for artificial intelligence without congressional authorization” and could displace state police powers in areas like consumer protection and civil rights. Members of Congress echoed this, calling the order an "end-run" around the legislative process and introducing legislation and oversight efforts to limit agency authority and prevent federal preemption of state AI laws absent clear statutory direction.
State Laws
The Big Picture
In 2025, state lawmakers in all 50 states introduced 1,208 AI-related bills. 145 were enacted into law. The most active areas fall under the category of consumer protection, generative AI, algorithmic accountability, employment and deepfake restrictions.
States continue to be hyper-active into 2026. According to one tracker, state lawmakers in 45 states have already introduced 1,561 AI-related bills (from January to March). If those numbers are accurate, 2026 is on pace to doubling the number of AI bills originating in the state legislatures. This number is already close to tenfold the number of bills passed in 2023.
Areas of concern
Tracking down or ranking the top concerns of states is difficult, and is somewhat subjective.
Many state legislatures have focused on the same concerns listed in the December Executive Order, signifying perhaps more of an alignment, not merely a patchwork, of state laws, despite the administration's earlier claims.
On the other hand, some of the bills listed below may be vulnerable to review by the forthcoming AI Litigation Task force, which the Trump administration has set up in order. to"challenge State AI laws inconsistent with the policy set forth in section 2" of the Executive Order on AI.
Below, we will highlight some of the major clusters of activity from the states, highlighting the underlying uniformity and possible areas of conflict that seem to be emerging in rough outline.
I: Algorithmic Accountability and Consumer Protection
26 states now regulate algorithmic discrimination, with a heavy concentration of those on employment and hiring practices.
By early 2026, New York City, Illinois, and Maryland had active laws requiring bias audits or consent for AI hiring tools, while broader consumer protection frameworks were pending in states like Washington and New Jersey.
Colorado SB 24-205 (The Colorado AI Act) is considered by some to be the first comprehensive risk-based state law in the U.S. It categorizes AI systems (e.g., "high-risk" vs. "general purpose") and mandates that deployers of high-risk systems—such as those used in lending, housing, or healthcare—complete detailed algorithmic impact assessments and notify consumers if AI is a substantial factor in a consequential decision
In Florida, CS/SB 482: Artificial Intelligence Bill of Rights includes three main provisions. First, government employees and entities are prohibited from extending, renewing, or entering into contracts with AI companies. Second, AI companies and their chatbot features must block minors from creating an account unless a parent or guardian gives express consent. Third, bot operators must periodically disclose to users that the interaction is automated, in order to prevent deception and ensure users understand the nature of the interaction.
This bill has not passed the necessary votes in the Florida Legislature. After a large push from Gov. DeSantis, the Florida House of Representatives has stalled the measure.
II: Transparency
Transparency appears to be one of the most consistent concerns. 45 states have introduced legislation focused on transparency.
Specific disclosure mandates (e.g. "I am a bot") are now active in at least 5 states, including Utah, California, and Illinois.
In Washington, HB 2225: Regulating AI Companion Chatbots requires AI companies and their chatbot operators to "provide a clear and conspicuous disclosure that an AI companion chatbot is artificially generated and not human." This disclosure must happen at the beginning of a user opening the chat and then once every three hours. Additionally, if the operator of the chatbot knows the user is minor, they must implement reasonable measures to prevent the generation of sexually explicit content and prohibit the use of manipulative engagement techniques, such as prompting users to return emotional support, providing excessive praise, or mimicking romantic relationships. It further seeks to protect minors from when a chatbot might veer into discussions of self-harm and suicide ideation.
In Virginia SB 796: Artificial Intelligence Chatbots and Minors Act requires AI companies and their affiliate chatbot software to implement reasonable systems that help ensure user, especially minors, safety when interaction with chatbots. It also requires the companies to make a reasonable effort to contact emergency services or law enforcement if it obtains information suggesting an imminent risk or threat to a user. Further, it requires a submitted report to the Attorney General after an incident is known.
California AB 2013 (Generative AI Training Data Transparency Act) law tackles the "black box" of AI development. It forces developers of generative AI systems to publicly post a high-level summary of the datasets used to train their models, including whether the data includes copyrighted works or personal information. Defenders say it shifts the focus from how AI is used to how AI is built.
Also in California: SB 53: Transparency in Frontier Artificial Intelligence Act has two purposes. First, it seeks to establish California as a leading innovator in artificial intelligence. This is to be done through a combined effort of public/private companies and the Joint California Policy Working Group on AI. These groups seek to "support effective artificial intelligence governance" and "balance the technology’s benefits and the potential for material risks." Second, it seeks to provide common sense laws to protect the public. When any AI Frontier Model (companies training large AI models, with over $500 million in annual revenue) make their models public, they are required to provide a disclosure describing how they will incorporate national safety standards, assess their models catastrophic risks, apply mitigations before deployment, identify and respond to critical safety incidents, and govern internal oversight. This report must be updated annually and published within 30 days. Of the bill, Governor Newsom said:
“California has proven that we can establish regulations to protect our communities while also ensuring that the growing AI industry continues to thrive. This legislation strikes that balance. AI is the new frontier in innovation, and California is not only here for it – but stands strong as a national leader by enacting the first-in-the-nation frontier AI safety legislation that builds public trust as this emerging technology rapidly evolves."
In Utah, HB 276: Digital Voyeurism Prevention Act requires AI companies and their operators to embed provenance data that would allow users to "determine if an image was created or altered through the use of AI." Provenance data is a type of data that is embedded into web content that allows users to verify the content's authenticity, origin, or modification history. These disclosers would help users of AI and of technology in general to get real-time verification of the authenticity of viewed content.
III: Privacy and Data Control
As of early 2026, 20 states have enacted comprehensive privacy laws that grant consumers specific controls over AI data. While 2024 focused on passing new laws, 2025 became the year of the "AI Amendment," with 8 states—including Colorado, Connecticut, and Virginia—updating existing privacy frameworks to specifically restrict how personal information is used to train generative AI models and requiring developers to disclose if user data is being "scraped."
In Utah, SB 256: Identity Protection Modifications adds additional provisions to Utah's existing libel and slander law. It clarifies that defamation law now applies to content created through AI technology and provides a route to recourse for persons or companies who bring suits of either libel and defamation.
The Minnesota Consumer Data Privacy Act (MCDPA), sets a national benchmark by giving citizens the "right to an explanation," forcing companies to reveal the specific data points that led an algorithm to make a negative decision about them. Some say it is the most protective AI privacy law in the U.S.
In Virginia, SB 85: Consumer Data Protection Act amends Virginia's Consumer Data Protection Act (VCDPA) to expand consumer rights over personal data. It requires interoperability between social media platforms and AI system operators. This amendment would allow consumers to access, correct, delete their personal data, delete social graph data, and delete contextual data. It would also grant consumers the right to opt out their data being used for personalized ads, the sale of their personal data, and further profiling.
Additionally, states are pushing into "biological privacy," with Montana leading the way by passing laws that treat neural data (brain-wave information) as a protected sensitive category, preventing AI companies from collecting mental data without explicit, high-level consent.
The Illinois Biometric Information Privacy Act (BIPA) requires companies to get a written release before collecting fingerprints or "faceprints." Because it allows individuals to sue for $1,000 to $5,000 per violation, it has forced AI giants like Google and Meta to pay billions for unauthorized facial recognition training.
IV: Promoting Innovation
3 states (Utah, Texas, Delaware) have formally established AI "Regulatory Sandboxes" that allow companies to test products with temporary waivers from certain state regulations.
Utah SB 149 (Artificial Intelligence Amendments) also creates an "AI Learning Laboratory." It allows AI companies to apply for a "safe harbor" to test innovative applications under state supervision for up to 24 months. If the technology proves safe and beneficial, the state can use the findings to permanently adapt its regulations, effectively letting startups help write the rules.
For a full list of all the AI and Data Privacy bills proposed or passed this year, see this site.
Future of a Federal Guideline
The much criticized "patchwork" of state AI laws is, in practice, producing a surprising degree of uniformity. But this might not be as surprising at one might think. Historically, states have taken the reins on countless new technologies, regulatory schemes, and the like, and have built public trust from the ground up.
Take, for example, the famous "laboratories of democracies" line from Justice Brandies in New Ice State Co. v. Liebmann.
"To stay experimentation in things social and economic is a grave responsibility. Denial of the right to experiment may be fraught with serious consequences to the nation. It is one of the happy incidents of the federal system that a single courageous State may, if its citizens choose, serve as a laboratory; and try novel social and economic experiments without risk to the rest of the country."
The Role of the States
In a hearing on September 18, 2025, Law Professor Neil Richards testified in front of the House Judiciary Committee on the future of AI regulation. As part of his message, Professor Richards highlighted the role states have played over the past three decades to address emerging technologies. He said, "If states had been banned from regulating the internet, there would be no mandatory privacy policies in e-commerce, no data breach notification requirements, no laws preventing employers from demanding employees’ social media passwords, no restrictions on facial recognition without consent, and no comprehensive privacy statutes at the state level." He pointed to California's Consumer Privacy Act, which became the model for other states and eventually the federal government's data privacy statutes. He ends by saying, "Congress faces a stark choice. It can deny 'the right to experiment' that exists within the American federal system through premature preemption, or it can allow states to continue their crucial role as laboratories of “social and economic experiments” in regulating technology."
Professor Richards is not alone in his belief about the power of states to find innovate and collaborative solutions. A letter signed by over 50 state senators, called for President Trump to allows states to legislate on AI and to "engage with us as partners in shaping a framework that protects Americans and strengthens U.S. leadership."
Concluding thoughts
Even though AI regulation is still in its infancy, the volume and uniformity of laws suggest a clustering along a few main fronts, including algorithmic and consumer protection, transparency, privacy, and innovation.
When states are given their full capacity to work through tough problems, to consult their constituents, and to experiment, we stand a better chance resisting some of the major national, economic, and social challenges posed by AI.
While there may be a need for a federal regulatory scheme in the future, our present situation is giving states the opportunity to do what they do best: exercise their constitutional authority to respond to emerging challenges. As Madison observed, “The State governments, with the people on their side, would be able to repel the danger.”
For more on AI regulation and the big questions about state's frameworks, see:
Co-Governance and the Future of AI Regulation — Harvard Law Review
National Legislative Policy Framework for AI — National Governor's Association
Introduction
It was been quite some time since our newsletter covered the Supreme Court. Thankfully, April has been full of interesting and important cases that center around state authority, federal preemption, and federalism.
Case 1: Trump v. Barbara (Birthright Citizenship)
Background: On January 20, 2026, President Trump signed Executive Order 14160: Protecting the Meaning and Value of American Citizenship. This EO sought to limit the traditional guarantees of the 14th Amendment's Citizenship Clause. Under its order, persons born in the United States would not be automatically granted citizenship if either of their parents were illegal or temporary residents at the time of birth. After being blocked by multiple courts, the Government sought a petition for review. In this, they said that the citizenship clause was "adopted to confer citizenship on the newly freed slaves and their children, not on the children of aliens temporarily visiting the United States or of illegal aliens.” The Court granted certiorari and scheduled oral arguments for April 1.
Procedural History: Following the order's publication, it prompted a wave of litigation. In Seattle and Maryland, federal judges issued preliminary injunctions on the order. Solicitor General Sauer sought a petition at the Supreme Court, asking the Justices to weigh in only on the legitimacy of preliminary and universal injunctions. In this case, Trump v. CASA Inc., the Court held that universal injunctions likely exceed the power Congress granted to federal courts. In Barbara v. Trump, a federal judge once again blocked the order with a preliminary injunction and a divided panel of the 9th Circuit Court held, in Trump v. Washington, that the order is invalid "because it contradicts the plain language of the Fourteenth Amendment’s grant of citizenship." These cases were consolidated into Trump v. Barbara.
Oral Argument: At argument, the government leaned heavily on a domicile theory, claiming that the Citizenship Clause excludes children of undocumented immigrants because their parents lack a “legally recognized domicile” in the United States. General Sauer argued that Wong Kim Ark supposedly turned on the fact that Wong’s parents were lawful, permanently domiciled residents, and that this implied domicile is a constitutional prerequisite for birthright citizenship. Cecilia Wang countered that this reading distorts Wong Kim Ark, which held that virtually all children born on U.S. soil are citizens, regardless of their parents’ immigration status. She emphasized that the decision never made domicile a constitutional requirement and that the government was trying to smuggle in a limitation the Court had explicitly rejected. Wang also stressed that undocumented immigrants are unquestionably “subject to the jurisdiction” of the United States, and that the government’s domicile theory would allow the executive branch to narrow a constitutional guarantee without congressional action or historical support.
Federalism Implications
A ruling upholding the order would mark a significant shift in federalism by allowing a federal statute to override a primarily shared function between the federal government and states. States rely on a status quo of citizenship to perform a variety of functions and a change away from this status would reduce a state's authority over millions of people. States like California argue that defining who is a citizen within their borders is essential to public order. In a 2026 amicus brief, a coalition of Attorneys General stated, "The federal executive cannot unilaterally reach into the delivery rooms of our states to categorize our residents into tiers of belonging, destabilizing the very fabric of our communities."
A ruling striking down the order, by contrast, would uphold the long-standing tradition of territorial jurisdiction, something that states exercise in numerous ways.
Case 2: Chiles v. Salazar (Conversion Therapy)
Background: In 2019, Colorado passed a law prohibiting licensed counselors and therapists from engaging in "conversion therapy" with minors. The law defined conversion therapy as “any practice or treatment... that attempts to... change an individual’s sexual orientation or gender identity," as well as any "effort to change behaviors or gender expressions or to eliminate or reduce sexual or romantic attractions toward individuals of the same sex." The law does allow for therapists to provide "acceptance, support, and understanding for... identity exploration and development." Ms. Chiles, a licensed therapist in Colorado, filed a suit in federal court claiming that the law unconstitutionally reached into her talk therapy practice and was a violation of her First Amendment rights.
Procedural History: At the district court, Chiles argued that her practice does not try to "convert" the clients, but rather only tries to help them with their stated desires and goals in counseling. She neither affirms or denies a client's sexual attractions or feelings, but works within their specified goals to help eliminate or bring into harmony, these feelings. The district court denied rejected Chiles' request for a block on the law as her challenge went through courts. The 10th Circuit court agreed with the district court. In their opinion, they held that the state's ban on conversion therapy regulated conduct and that speech is incidentally apart of this conduct. As such, the law should be analyzed under a rational basis test, the lowest form of judicial review. This test requires a state to have a legitimate interest in which the law/statute protects and that there is a rational connection between its means and goals. Chiles challenged these lawsuits in a writ of certiorari. The Court granted this and heard arguments in March.
Decision: In an 8-1 decision, the Court held that the lower court should have used a more stringent standard, such as strict scrutiny, to determine whether or not the law passed First Amendment standards. Additionally, the Court noted that the law was not likely to pass this test, because it "censors speech based on viewpoint." Justice Gorsuch also noted that the First Amendment “reflects... a judgment that every American possesses an inalienable right to think and speak freely, and a faith in the free marketplace of ideas as the best means for discovering truth" He added, "any law that suppresses speech based on viewpoint represents an egregious assault on both of those commitments.”
Federalism Implications:
The reactions to this decision have been mixed. Some in support believe this is a step in the right direction and a sign of the Court's increasingly pro-First Amendment jurisprudence. Others, have criticized this opinions for the effects in may have down the line. While the ruling specifically addresses Colorado’s law, it directly threatens similar bans in over 20 states and dozens of municipalities, as counselors in those jurisdictions can now file similar First Amendment challenges.
In an article, Erwin Chemerinsky, a law professor as UC Berkeley, highlighted some of these effects. He points out past Supreme Court cases that have dealt with First Amendment challenges to laws that regulate "professional speech" and how these cases have had inconsistent results. He noted, "Besides being inconsistent with the court’s own precedent, the decision in Chiles v. Salazar also has some deeply troubling implications which the majority makes no attempt to address. Based on this decision, can the states effectively regulate any professional speech?" This was one of the primary concerns of the dissent, given by Justice Jackson. She explained, "Speech uttered for purposes of providing medical treatment may be restricted incidentally when the State reasonably regulates the speaker’s provision of medical treatments to patients.” Chemerinsky highlights this part of Justice Jackson's opinion as a way to explaining the limiting rule this could enforce on states.
In response to the ruling, Colorado legislators are already advancing new bills, such as the Civil Actions for Conversion Therapy Survivors , which would allow individuals to sue practitioners for damages independently of the now-weakened state ban
For the full text of Dr. Chemerinsky's article, see this link.
For more on the debate, see this "Courthouse Steps" panel, provided by the Federalist Society
Case 3: Watson v. Republican National Committee (Election Law)
Watson v. Republican National Committee is currently pending before the Supreme Court of the United States, making it one of the most consequential election law cases of the 2025–2026 term. The Court granted certiorari on November 10, 2025, and heard oral argument on March 23, 2026; as of now, no decision has been issued, and a ruling is expected by the end of the term, likely in June 2026.
Background: Mississippi law, along with laws from dozens of other states, allows for mail-in-ballots to be counted as long as they are received within five business days after Election Day (the Tuesday next after the first Monday in November). In 2024, the Republican National Committee, along with the Mississippi Republican party, challenged the law. They argued that federal election law sets a national standard on when votes and ballot should arrive. After two appeals, the Supreme Court granted certiorari and heard arguments on March 23.
Procedural History: At the district court level, Judge Guirola upheld the law. He argued that Congress's intent with the national election day statute was to eliminate the administrative burden and to risk the chance that a state could have an undue influence on another by setting their election day earlier. On appeal, the 5th Circuit reversed, and held that the federal statute applies to both ballots cast and received by election officials.
Oral Argument: At oral argument, Mississippi Solicitor General Stewart argued that states have traditionally held the power to direct and administer elections, as long as they are consistent with federal law. Furthermore, he argued that Mississippi's law was consistent because voters make their final decisions by or on election day. Mr. Clement, who argued on behalf of the challengers, argued that when Congress passed the statute, "the casting of ballots and the state’s receipt of ballots were 'so inextricably intertwined' that no one would have thought of one without the other." Some of the Justices brought up concerns about recall votes and what that means for elections. Others, like Justice Alito, was concerned that Mississippi's law, taken to an extreme, could leave room for other states to set unrealistic timelines for their own ballots to be counted. A 2022 law, the Electoral Count Reform Act, was a key point in the argument as well. This law refers to a "period of voting" which phrasing implies that Congress was fine with states have a period of time rather than a specific day of voting, according to Justice Kagan.
Federalism Implications
The case centers on a deceptively simple but legally significant question: whether federal law requires that ballots be received by Election Day, or whether states retain the authority to count ballots that arrive after Election Day so long as they were mailed on time.
The stakes are viewed as relatively "high" because the Court’s decision could reshape election administration across the country. If the Court affirms the Fifth Circuit, it would likely invalidate ballot grace periods used in many states, effectively imposing a nationwide rule that ballots must be received by Election Day. If, by contrast, the Court reverses, it would preserve state discretion to define when a ballot counts as “cast,” reinforcing traditional state authority over election administration.
Oral arguments suggested a divided Court, with some justices appearing receptive to a stricter, receipt-based interpretation of federal law, while others expressed concern about overriding longstanding state practices.
Concluding thoughts
In February, we highlighted the variety of redistricting efforts nationwide. Cases like Watson v. RNC are another example of a new pattern, with states stepping in to fill the legislative void created by Congress, and then the federal government simply stepping back into the picture to regulate the process once again.
This type of phenomena has been studied before and is known as "Post-Structural Federalism." An article by the Cato Institute speaks to this new wave of federalism.
"Traditional theories of federalism presume a functioning federal government. Cooperative federalism assumes collaboration. Uncooperative federalism assumes resistance. But both require a federal partner that still governs. Post-supremacy federalism begins from a different premise: The federal partner has abrogated its implicit contract with the states, which now wield contingent sovereignty. They act without formal delegation and govern under the constant threat of reversal—from preemption, defunding, or snapback enforcement. States may be the laboratories of democracy, but the federal government can step in at any time to overturn the experiments, using laws already on the federal books."
Whether this new pattern is healthy or not may come down to the political environment of the day. As with most give-and-take relationships, federalism works when both partners in the arrangement exercise their maximum power within Constitutional bounds.
Other Interesting Cases:
Louisiana v. Callais (Voting Rights Act): In a 6-3 decision, the Court held that Louisiana’s SB 8 map violated the Equal Protection Clause because its race-based districting was not justified by a compelling interest, as §2 of the Voting Rights Act did not actually require the creation of an additional majority-minority district. Writing for the majority, Justice Alito said, "Section 2 of the Voting Rights Act of 1965, was designed to enforce the Constitution— not collide with it. Unfortunately, lower courts have sometimes applied this Court’s §2 precedents in a way that forces States to engage in the very race-based discrimination that the Constitution forbids." This could have sweeping impacts of dozens of districts created primarily on race-based considerations. This case was decided April 29.
Little v. Hecox (Transgender Athletes): Argued in December, with a decision pending. The Court is deciding "whether laws that seek to protect women's and girls' sports by limiting participation to women and girls based on sex violate the equal protection clause of the 14th Amendment."
West Virginia v. B. P. J.: Argued in December, with a decision pending. The Court is deciding whether Title IX prevents a state from designating girls' and boys' sports teams based on biological sex and whether the equal protection clause of the 14th Amendment is violated when state offers separate boys' and girls' sports.
Comments