The year 2024 was supposed to be the “AI apocalypse” for elections, a time when fake content would run rampant and sway voters. Yet, that widespread catastrophe didn’t quite materialize. We saw some attempts, like the robocall mimicking President Biden’s voice in New Hampshire, urging people to delay their vote – but these fakes often had a limited impact. There were also rather obvious tells for AI-generated visuals, like a person in a photo having too many or too few fingers, or short videos shot from a single, unchanging angle, making them easier to spot as artificial. Back then, political campaigns were also cautious; they worried about public backlash if they openly used AI tools that most voters didn’t understand, or if they were caught secretly using AI-generated content. It was a time of unease, but the worst fears hadn’t yet come true.
Fast forward to today, and the landscape has drastically shifted. The democratic norms we once relied on have eroded, and the dangers of AI-generated fakes are far more potent, while public outrage seems less of a deterrent. We’re seeing alarming incidents, like a White House account in January posting an altered photograph of an African American protester, her skin darkened and tears added, to manipulate perception. Anti-ICE activists have also engaged in “unmasking” immigration officers with AI-generated images, a risky tactic given AI’s tendency to “hallucinate” facial features and details, creating potentially false accusations. Looking ahead to the 2026 midterm elections, the National Republican Senatorial Committee released a deepfake video of a Democratic Senate candidate, James Talarico, appearing to “read” his own old social media posts. Although a tiny disclaimer “AI Generated” flickers in the corner, many voters might miss it, believing the politician actually said those words, fundamentally blurring the lines of truth.
The heartbreaking truth is that federal action to curb this AI-driven deception seems unlikely. The previous Trump administration took a firm stance against any AI regulation, prioritizing innovation over safeguards. In March, then-President Trump explicitly called for removing “outdated or unnecessary barriers to innovation” to speed up AI deployment. While some Democrats in Congress, like Senator Brian Schatz and Representative Don Beyer, have pushed back with proposals to counter this approach, and Senator Mark Warner has suggested various measures for social media companies to tackle “maliciously manipulated media,” the overall resistance to regulation casts a long shadow over effective federal intervention. This lack of a unified, proactive federal stance leaves states and individual citizens to grapple with the growing tidal wave of AI-generated misinformation.
Despite the dim prospects for federal oversight, the problem isn’t going unaddressed at all levels. More than two dozen states have introduced laws concerning deepfaked political content during elections. However, most of these laws merely mandate disclosure of AI use rather than prohibiting it. Critics, like Sarah Kreps from the Cornell Brooks School Tech Policy Institute, point out a fundamental flaw: “Legislation is a noble effort, but the technology is moving so fast. You are going to be addressing yesterday’s problems.” The pace of legislative action simply can’t keep up with the exponential evolution of AI capabilities. This technological sprint means that by the time a law is enacted, the AI landscape it was designed to regulate has likely already transformed, rendering the legislation partially, if not entirely, obsolete.
Given the inadequacy of federal and state-level legislative responses, the burden of combating AI-driven misinformation increasingly falls on social media platforms and individuals. Tim Harper, from the Center for Democracy and Technology, stresses the importance of platforms recommitting to their 2024 election safety pledges. This means actively raising public awareness about deepfakes, equipping voters with the skills to identify AI-generated content, and transparently communicating their efforts to detect and remove such material. Harper also suggests that political campaigns themselves could play a crucial role by “invest[ing] heavily in using content provenance—watermarking any of their authentic press releases and videos and images—not only to give a trust signal to voters… but also to prevent the risk that they would be deepfaked.” This dual approach, combining platform responsibility with campaign-level authentication, could create a more resilient information environment.
Ultimately, the fight against AI-generated misinformation hinges on media literacy and critical thinking. Organizations like Factchequeado, a fact-checking and media literacy network, are already on the front lines, educating Latino voters about deepfakes and the wider world of mis/disinformation. They provide reliable Spanish-language news and run a WhatsApp chatbot that allows users to submit claims for verification. Laura Zommer, Factchequeado’s co-founder, encourages voters to develop a habit of consulting trusted sources and to continuously train their “eye to look for details that can show you a clue that it is not necessarily true or authentic content.” Similarly, the Poynter Institute, through its MediaWise initiative, empowers people to critically examine content. They’ve developed resources like instructional videos for seniors on reverse image searching and an “AI Unlocked” toolkit for teens, focusing on visually recognizing AI-generated materials and using AI responsibly. Sean Marcus, an interactive learning designer at MediaWise, advises vigilance: “expect to see more and more extreme misinformation, twisted information, and out-of-context information.” Yet, he warns against fatalism, emphasizing that we don’t have to passively accept the flood of misinformation. We, as individuals and as a society, have the power to act, to sharpen our discernment, and to recognize the good information from the bad, ensuring that truth, not deception, shapes our collective future.