One photo of President Donald Trump leaning on a walker has been seen by millions of people online-but it never happened. The AI-generated image was tweeted by Democratic strategist Keith Edwards with a pointed jab at Trump’s newly signed executive order restricting states from regulating AI. Though many users quickly flagged it as fake, its rapid spread serves only to illuminate the growing potency of synthetic media to shape political narratives and lend new urgency to policy debates over AI governance.
Scholars have long warned that political deepfakes operate on a spectrum of intent and realism. The Trump walker image, as found at Purdue University’s Governance and Responsible AI Lab, would fall under the typology as a “foefake”-unrealistic and negative, intended more for satire than deception. While such content may not persuade viewers of its literal veracity, it shapes perception through humor, cultural reference, and emotional framing. It is unsurprising, then, that in the 2024 presidential election, 78.3 percent of the deepfakes detected were unrealistic. Foefakes and fanfakes dominated over the realistic so-called “darkfakes” designed to mislead. That makes all the difference when it comes to strategies for detecting them-foefakes are less about technical trickery and more about rhetorical impact.
If satire were the goal, it was spot on. But the executive order at the core of this controversy is anything but satire. Dated December 11, 2025, it marshals federal agencies to challenge state AI laws, citing the need to avoid a “patchwork” of rules that slow innovation. Trump had said “one central source of approval” is what is needed, threatening that corporations facing 50 separate state regimes “could forget it.” The order commands the Department of Justice to create an AI Litigation Task Force, the Commerce Department to identify “onerous” state laws within 90 days, and the Federal Trade Commission to issue a policy statement on when state mandates to alter “truthful outputs” of AI models are preempted by federal law.
Funding arguably constitutes one of the most aggressive levers of the order. The states whose laws are considered in conflict with federal AI policy risk losing access to BEAD-program dollars-a blow that could reach $1.8 billion in California and $405.9 million in Colorado. Moreover, the order instructs agencies to review other discretionary grants for similar restrictions, essentially linking infrastructure and development funds with a state’s AI regulatory posture. So California and Colorado are squarely in the crosshairs. California has passed laws that require transparency for AI, ban certain algorithmic pricing practices, and include mandates for public tools to spot AI-generated media.
Colorado’s law addresses algorithmic discrimination and disclosure in consequential decision-making systems. The administration has targeted such measures as forcing “false results” or embedding “ideological bias” into the AI models. Supporters of the laws say they are necessary to protect against harms ranging from discriminatory hiring to misleading deepfakes. The deepfake problem is decidedly not theoretical in this policy fight. California’s transparency rules include provisions mandating that AI makers supply identification tools for synthetic media-precisely the kind of mechanism that could flag images like the Trump walker foefake before they go viral.
In challenging such laws, the executive order risks weakening defenses against political misinformation at the very moment it is proliferating. This is also a historical point of tension: federal versus state authority in new technologies. Through telecommunications, environmental regulation, and data privacy, states have frequently moved to impose stricter standards than federal law prescribes. Here, the administration is claiming priority-as it did not in previous executive orders-without explicit congressional consent.
Legal scholars say that decision may be fraught under separation of powers and the Spending Clause. One critic lamented that the order “removes” precautions without their replacement, leaving citizens vulnerable in service of industry flexibility. Tech industry leaders-including venture capital firms and AI developers-have welcomed the order as a step toward a unified national framework. The divergent rules enacted at the state level impede deployment and erode America’s competitive edge against rivals, especially China, they said. Proponents even acknowledge, however, that the executive order can’t fully pre-empt state law; lasting change would require legislation.
An executive order to create a federal framework on AI, with carve-outs for child safety and state procurement, was an indication of intent to legislate – though Congress has yet to pass any comprehensive AI regulation, despite introducing more than 150 bills in the last session. The Trump walker deepfake will disappear from feeds, but the collision it represents-between synthetic political imagery and the fight over who controls AI-is heating up. In the absence of any clear federal standard, states have filled the vacuum with laws targeting everything from bias to misinformation. Now, with those laws about to be challenged by federal agencies and funding on the line, the stakes go far beyond satire.
