Ron DeSantis Says an Executive Order Cannot Preempt State AI Laws

For Florida counsel, the preemption question is practical rather than theatrical: as of Q3 2026, Florida’s enacted AI-related obligations should not be treated as erased by EO 14365. DeSantis’s June 11, 2026 formulation — “an executive order can’t block states” — is directionally right for the compliance question, but it matters because it aligns with the operative hierarchy: an executive order may direct federal enforcement priorities and litigation posture; it does not, by itself, repeal or preempt state law. [1]

The order itself is also narrower than much of the political coverage suggests. Section 8(b)(ii) carves out state consumer protection measures from the order’s preemption targets, which is a material point for Florida exposure because the state’s live AI-adjacent enactment, SB 484, sits in the consumer-and-ratepayer protection lane rather than in the failed AI Bill of Rights lane. [2]

This is a risk read, not legal advice. The immediate working answer is that Florida’s enacted law and court-rule obligations remain part of the compliance inventory unless a court, Congress, or a competent implementing authority with valid statutory footing changes the analysis.

The operative stack for Florida exposure

That table is the cleanest way to avoid the common error. Florida has more than one AI-related development, and they do not have the same legal status. SB 484 is enacted. Rule 2.515(d)(2) is an effective court rule. SB 482 is dead. EO 14365 is federal executive direction. Those categories should not be collapsed into one headline about state AI law being “blocked.”

Why the executive order does not do the preemption work by itself

Preemption analysis starts above the executive order. The Supremacy Clause gives controlling force to the Constitution, federal statutes, and valid federal law made under constitutional authority. A presidential order can organize the executive branch and direct federal litigation priorities. It does not become a congressional preemption statute merely because it announces a national preference for less state-by-state AI regulation.

That distinction matters in Florida because the order’s strongest practical effect is likely to be agenda-setting: DOJ can select targets, frame theories, coordinate an AI Litigation Task Force, and pressure states through federal-program conditions where agencies have implementing authority. The order’s BEAD-program restrictions belong in that latter category: they may affect funding administration, but they depend on agency implementation and remain vulnerable to challenge. [2]

The failed federal preemption efforts in 2025 are therefore not trivia. If Congress had enacted a clear AI-preemption provision, Florida counsel would be reading statutory text, savings clauses, effective dates, and retroactivity language. Instead, practitioners are reading an executive order and anticipated federal litigation. That is a different risk posture. Latham & Watkins, assessing the order shortly after issuance, concluded that its arguments “do not seem to provide an obvious path to victory in the courts.” [3]

Section 8(b)(ii) narrows the order further. The carve-out for state consumer protection measures means the order does not even purport to attack every state law touching AI. That language is particularly relevant for SB 484 because the Florida enactment described in the current materials concerns data-center ratepayer protections, not a comprehensive AI-governance code or a state-level model-registration system. [2][4]

What Florida obligations remain live

SB 484 remains in the compliance inventory

SB 484 was signed on May 7, 2026. The available record identifies it as a data-center ratepayer protection measure. That is enough to make the Section 8(b)(ii) carve-out operationally important, but not enough to support sweeping claims about all AI systems, all platforms, or all developers. The narrower conclusion is the safer one: parties with Florida exposure in data-center development, power procurement, utility-cost allocation, or ratepayer-impact planning should keep SB 484 in the live-law column. [4]

For in-house teams, the consequence is less glamorous than the national fight. The controls that matter are document retention, regulatory ownership, contract review for data-center commitments, and escalation triggers when a project could shift infrastructure costs onto Florida customers. EO 14365 may be relevant if federal litigation later targets the statute, but that is not the same as a current compliance exemption.

Rule 2.515(d)(2) is a separate court-rule problem

Florida Supreme Court Rule 2.515(d)(2) took effect June 15, 2026, creating a statewide AI certification standard. [5] That obligation should be tracked separately from state legislation. Court rules operate through the judicial system’s authority over filings and practice; they are not the same instrument as SB 484, and they do not rise or fall simply because the White House dislikes state AI regulation.

The practical burden lands on lawyers, legal departments, and litigation-support teams. Someone has to know whether AI-assisted work triggers the certification requirement. Someone has to preserve enough process history to support a certification if a filing is challenged. Someone has to train outside counsel and vendors not to treat generative-AI review as an invisible drafting shortcut. Those are court-facing controls, not public-policy commentary.

SB 482 should not be treated as law

The proposed Florida AI Bill of Rights, SB 482, died in the House on March 13, 2026. [6] Its Senate vote history has been reported with some procedural variation, but the vote-count discrepancy is not the compliance point. The bill did not become law.

Speaker Perez’s alignment with the White House preemption strategy matters because it helps explain why the bill did not advance. It does not create a substitute obligation. Risk teams should be careful here: proposed rights language, legislative findings, and advocacy summaries can be useful for horizon scanning, but they should not be inserted into a control matrix as if they were enacted Florida duties.

The OpenAI suit is state enforcement, not preemption proof

Attorney General James Uthmeier’s June 1, 2026 civil lawsuit against OpenAI is also a separate bucket. [7] It may tell companies something about Florida’s enforcement appetite, but it does not answer whether EO 14365 can displace Florida statutes or court rules. A live state enforcement action can proceed on its own footing unless and until a court says otherwise.

The dormant commerce clause theory is serious, but unsettled

The most plausible federal attack is not magic executive-order preemption. It is litigation arguing that certain state AI laws impermissibly burden interstate commerce. That theory is not frivolous. AI products are often deployed across state lines, and a state-by-state compliance patchwork can impose real operational burdens. The harder question is whether those burdens are legally excessive under current doctrine.

No court has yet ruled on a dormant commerce clause challenge to a state AI law under this framework. That absence should restrain both sides. It is too early to say the White House theory is dead on arrival, and equally too early to tell Florida clients that state AI laws have already been neutralized.

The first weakness is technological. The order’s theory draws on an older internet-indivisibility model associated with the Pataki era: the idea that online services cannot realistically comply state by state. A January 2026 Harvard Law Review student note argues that this model is undercut by more recent Supreme Court treatment of geolocation and compliance toggling, including Free Speech Coalition v. Paxton. The source is useful as legal analysis, but it should be identified for what it is: a student note, not peer-reviewed scholarship or binding authority. [8]

The second weakness is extraterritoriality. National Pork Producers v. Ross rejected the idea that out-of-state economic effects alone invalidate a state law. That does not give states unlimited power to regulate national technology markets. It does mean a plaintiff needs more than the fact that a Florida rule changes conduct by a company operating outside Florida. [8]

The third weakness is evidentiary. Pike balancing asks whether the burden on interstate commerce is clearly excessive in relation to the local benefits. A complaint can allege patchwork burden, engineering difficulty, and chilled deployment. But the balancing test ultimately needs evidence. The executive order assumes a burden profile; it does not, by issuing the assumption, supply the record a court would need. [3][8]

That is where practitioners should resist overstatement. Dormant commerce clause litigation could eventually narrow a state AI law, enjoin an especially burdensome provision, or produce a record that changes the risk calculation. It has not yet done so for Florida’s enacted SB 484 or the Florida Supreme Court’s AI certification rule.

A Florida-facing risk position for Monday morning

The cleanest Monday-morning position is to separate five buckets before deciding what controls remain in place.

• Political pressure: EO 14365 and White House messaging may shape DOJ priorities and state legislative behavior, but they are not self-executing repeal instruments.
• Federal statutory law: no current congressional AI-preemption statute has been identified as in force, so counsel should not brief federal preemption as already accomplished.
• Enacted Florida law: SB 484 remains live and is especially hard to dismiss under the executive order because of the Section 8(b)(ii) consumer-protection carve-out.
• Florida court rules: Rule 2.515(d)(2) is an active judicial-practice obligation and should be owned by litigation, outside-counsel management, and filing-review teams.
• Predictive litigation: dormant commerce clause challenges may come, but current analysis is predictive rather than precedential.

For a partner briefing a client with Florida exposure, that means AI-related controls tied to live Florida obligations should stay in place. SB 482 should come out of the live-obligations list because it died. SB 484 should stay in. Rule 2.515(d)(2) should stay in. The OpenAI suit should be tracked as enforcement context. EO 14365 should be tracked as a federal litigation and agency-implementation signal, not as a blanket permission slip to ignore Florida law.

References

1. DeSantis Interview on State AI Laws, Politico, June 11, 2026.
2. Executive Order 14365, The White House, December 2025.
3. Trump Administration Directs DOJ to Challenge State AI Laws, Latham & Watkins, December 2025.
4. SB 484, The Florida Senate, May 7, 2026.
5. In re: Amendments to Florida Rule of General Practice and Judicial Administration 2.515, Florida Supreme Court, effective June 15, 2026.
6. SB 482, Florida Legislature, March 13, 2026.
7. Attorney General Uthmeier Files Civil Lawsuit Against OpenAI, Florida Office of the Attorney General, June 1, 2026.
8. Executive Preemption After Pataki and Paxton, Harvard Law Review Blog, January 2026.