On December 11, 2025, the Trump Administration released the “Ensuring a National Policy Framework for Artificial Intelligence” Executive Order (EO), which clarifies federal policy related to artificial intelligence (AI) and calls for federal legislation to create a minimally burdensome national policy framework for AI (National AI Policy) that preempts inconsistent state law. The EO represents but one of several actions taken by the Trump Administration, including a July 2025 AI Action Plan and the launch of an AI “Genesis Mission” aimed at accelerating the application of AI in scientific discovery.
The EO creates a framework for federal agencies – including the Department of Commerce, Department of Justice, and Executive Office of the President – to assess and challenge the “most onerous and excessive” state laws “that threaten to stymie innovation” on the basis that they are preempted by federal regulation, impermissibly regulate interstate commerce, or are otherwise inconsistent with the US Constitution. The EO cites, as one example, a recent Colorado law banning algorithmic discrimination, claiming that it may force AI models to produce false results in order to avoid a ‘differential treatment or impact’ on protected groups. In response, states have largely refuted the prospect of federal preemption of state AI measures and have warned of “disastrous consequences” if the federal government were to dismantle state AI regulation or impose a state-level moratorium on AI regulation. For additional analysis of the EO, see our Legal Briefing.
Although the EO charts a course to potentially dismantle state AI regulations applicable to virtually any industry that can be regulated by the federal government, the impact of the EO on the insurance industry deserves special attention by virtue of the fact that, under federal law, the regulation of the business of insurance is largely delegated to the states. For their part, state insurance regulators have spent considerable time assessing how best to regulate the industry’s use of AI and, like their respective governors and attorneys general, are largely opposed to the policy outlined in the EO. For example, days after the EO was published, the National Association of Insurance Commissioners (NAIC) issued a statement expressing its “deep concern” with the EO and noting that state insurance regulators are already taking proactive steps to address AI challenges, including an AI Model Bulletin that was adopted by the NAIC in 2023 and has since been implemented in 25 states. The National Council of Insurance Legislators (NCOIL) issued a similarly dismissive press release expressing that its members were “greatly disturbed” by the EO. Current AI-related workstreams underway at the NAIC and NCOIL – including the NAIC’s development of an AI Systems Evaluation Tool – continue apace as if there will be no federal preemption.
Eighty years ago, Congress passed the McCarran-Ferguson Act (McCarran Ferguson)1 to establish that the “continued regulation and taxation by the several States of the business of insurance is in the public interest” and, subject to limited exceptions (including acts of “boycott, coercion, or intimidation”2), that:
No Act of Congress shall be construed to invalidate, impair, or supersede any law enacted by any State for the purpose of regulating the business of insurance, or which imposes a fee or tax upon such business, unless such Act specifically relates to the business of insurance….3
As such, pursuant to the express terms of federal statute, the federal government is generally only permitted to preempt state laws relating to the business of insurance by enacting a federal statute that specifically relates to the business of insurance. In that respect, the EO contemplates the need for Congressional action on AI regulation to “ensure a minimally burdensome national standard.” However, given the Senate’s recent 99-1 vote in opposition to a moratorium on state AI regulation, the intricacies of developing (and passing) an AI regulatory bill through Congress, and the need for the law to expressly relate to the business of insurance, it would be fair to conclude that (a) Congressional action impacting insurer AI usage – and state insurance laws regulating it – is unlikely in the near term and, (b) McCarran Ferguson will otherwise serve to block any federal attempts to challenge state insurance laws that are inconsistent with current federal AI policy.
The analysis is not so simple. As the Supreme Court has noted, “insurance companies may do many things which are subject to paramount federal regulation; only when they are engaged in the ‘business of insurance’ does [McCarran Ferguson] apply” (emphasis added).4 In the context of AI, the insurance industry – like all industries – has deployed AI systems across several business functions, not just those that exclusively relate to the business of insurance. And, although state insurance laws generally pertain to the regulation of the business of insurance, there may well be instances where the application and effect of those laws extend beyond the business of insurance.
To that end, the Supreme Court has defined “business of insurance” such that the reverse-preemptive effect of McCarran-Ferguson applies only in instances where:
- The practice has the effect of transferring or spreading a policyholder’s risk;
- The practice is an integral part of the policy relationship between the insurer and the insured; and
- The practice is limited to entities within the insurance industry.5
None of these criteria is dispositive in itself. Applying these factors, courts have found that activities which are directly tied to ratemaking and other functions at the core of and unique to the insurance industry – often including rating, underwriting, rate-making and form-making, claims-related functions, and certain agent/insurer and insurer/policyholder relationships (core insurance functions) – are generally within the scope of McCarran Ferguson and consequently immune from federal antitrust law.6 By contrast, courts have found several activities to not be the “business of insurance” even where those activities are conducted by an insurer or other insurance-regulated entity. Those activities include merger and acquisition functions, marketing practices, employment practices, and arrangements with third-party providers of non-insurance goods and services (non-core insurance functions).7
In that respect, the ability of the federal government to successfully challenge state insurance laws that it deems to be inconsistent with its current AI policy may turn on the specific application of the law and the specific purpose of the regulated AI activity, namely whether the state insurance law exclusively regulates the “business of insurance” or whether the regulated AI system is employed for use in a non-core insurance business function, the likes of which could include marketing, fraud detection, loss prevention, corporate governance, procurement, data processing, human resources, and other general corporate functions. Whether the federal government attempts such a tack (one that would be deeply unpopular with states), remains to be seen.
__________
1 15 U.S.C. §§ 1011- 1015.
2 15 U.S.C. § 1013(b).
3 15 U.S.C. § 1012(b).
4 SEC v. National Securities, Inc., 393 U.S. 453, 459-60 (1969).
5 See Union Labor Life Insurance Co. v. Pireno, 458 U.S. 119 (1982).
6 See e.g., Government Accountability Office, Legal Principles Defining the Scope of the Federal Antitrust Exemption for Insurance (March 4, 2005), available at: https://www.gao.gov/assets/b-304474.pdf, and the cases cited therein.
7 Id.
[View source.]
