Political deepfakes have marred elections worldwide. Deepfakes have
influenced
presidential elections in Argentina,
affected
parliamentary races in Slovakia, and
spread
misinformation widely.
In contrast, deceptive deepfakes have not swamped American politics. This is despite the easy availability of rapidly improving AI deepfake technology. It’s now easy to generate a high-quality deepfake that can deceive most viewers.
Generative AI has certainly entered American politics, and its use is set to expand dramatically in coming years. Politicians, notably including President Donald Trump, are widely using AI to generate pictures and videos that extol them and attack their adversaries, sometimes in extreme and offensive terms.
The nightmare scenarios of election integrity being undermined by pervasive deceptive deepfakes showing candidates saying or doing things they never said or did, or by late-breaking deepfakes that targets can’t counter, have not emerged.
Why haven’t we seen the nightmare scenarios unfold? It’s safe to say that America has not escaped political deepfake horrors because of a sudden outbreak of ethics among our politicians and their consultants.
Instead, tough state laws across the country have established new norms — backed by real consequences for violations — that stamped out an epidemic of deceptive deepfakes before it ever started. At the state level, examples show responsible government regulation of AI is possible and should be embraced,
Over the last three years, state lawmakers in red, blue, and purple states have passed laws making it a civil and criminal offense to disseminate deceptive deepfakes. These state laws typically require that a deepfake that purports to depict a candidate saying or doing something they did not say or do must be prominently labeled as “generated by artificial intelligence.” They don’t apply to satire or AI-generated imagery that isn’t realistic. As of July, these laws are now on the books in 31 states. Virtually every one of those laws passed with bipartisan support.
In one of the most notable deepfake incidents in the United States — a 2024 robocall in New Hampshire impersonating President Joe Biden — state prosecutors identified the perpetrator and quickly filed charges, sending a message that deepfakes won’t be tolerated.
There have been almost no comparable cases in the United States. While deepfake technology has been used in numerous advertisements and for social-media content, in virtually every case these are satire, exaggerated, and unrealistic — or properly labeled as AI-generated. For the most part, political deepfakes are not being used to deceive and defraud voters.
State anti-political deepfake laws are one of the first effective legislative responses to an AI-related harm. The harm emerged. Lawmakers recognized it, and, in a majority of states, they passed common-sense, narrowly tailored laws to address the danger. The laws have made a difference, not so much in punishing wrongdoers as in establishing societal standards and stopping abuses before they happen.
The Big Tech companies rushing increasingly powerful AI technologies to market insist that any efforts to impose safeguards will simultaneously fail and derail innovation. They are spending hundreds of millions of dollars on politics, primarily to block common-sense regulation, including a major effort to have the federal government preempt all state AI rules.
They tell us how dangerous and powerful the technologies they are unleashing are — but then ask us to trust them to solve the problem voluntarily.
The political deepfake experience rebuts Big Tech propaganda. State legislators identified common-sense solutions. They acted and got laws adopted. They have averted serious, foreseeable harm.
The lesson is that we the people do not need to passively accept whatever risks the Big Tech companies choose to impose on us, whether from deepfakes or data centers, chatbots threatening our kids, or AI that may make it easy to create bioweapons. “AI governance” must not mean how a handful of giant corporations choose to manage new technologies, but how our government places limits and guardrails on Big Tech’s deployment of AI.
Robert Weissman is co-president of
Public Citizen
(citizen.org), a progressive, nonprofit consumer-rights advocacy group and think tank based in Washington, D.C. Ilana Beller is director of state AI governance and tech policy at Public Citizen.
Robert Weissman
Ilana Beller
