What Are the Emerging Regulations Around AI Deepfakes in Elections?

The rapid advancement of generative artificial intelligence has made it increasingly easy to create highly realistic, fabricated audio, video, and images. When used in political contexts, these “deepfakes” pose a significant threat to electoral integrity by impersonating candidates, fabricating events, or spreading disinformation to manipulate voter behavior.

Following the major global election cycles of recent years, governments, regulatory bodies, and technology platforms have accelerated efforts to mitigate these risks. As of early 2026, the regulatory landscape surrounding political deepfakes relies on a combination of legislative bans, mandatory disclosure laws, and strict platform-level policies aimed at preserving transparency in democratic processes.

Legislative Approaches by Governments

Governments worldwide have adopted varying strategies to address synthetic media in elections, ranging from comprehensive AI frameworks to targeted election laws.

• The European Union AI Act: The EU has implemented some of the most comprehensive regulations globally. Under the AI Act, AI systems expected to pose significant threats to health, safety, or the fundamental rights of persons are classified as “high-risk.” The law also mandates that deepfake content must be clearly labeled as artificially generated or manipulated, ensuring voters are aware of its synthetic nature.
• United States State-Level Laws: In the absence of a singular, comprehensive federal AI law, numerous individual states have enacted their own legislation. As of January 2026, 28 states have enacted laws specifically addressing deepfakes in political communications. These laws typically require prominent disclaimers on any political advertisement featuring AI-generated content, restrict the use of such content within a specific window tied to the proximity of Election Day, or apply a combination of both approaches.
• Federal Regulatory Actions: Rather than implement new rules specifically addressing AI, the Federal Election Commission (FEC) voted in September 2024 to adopt an Interpretive Rule clarifying how its existing regulations on fraudulent misrepresentation apply to AI-generated content in campaign communications. Various federal legislative proposals have also been introduced to establish nationwide standards for watermarking and labeling synthetic political content.
• International Voluntary Accords: Several technology companies and democratic nations have signed voluntary accords committing to share detection tools and best practices to combat AI-driven disinformation in elections. One notable example is the Tech Accord to Combat Deceptive Use of AI in 2024 Elections, signed by 27 AI companies and social media platforms, which targets AI-generated content that deceptively fakes or alters the appearance, voice, or actions of political candidates and election officials.

Technology Platform Policies

Because social media platforms and search engines are the primary distribution channels for digital content, tech companies have implemented their own frameworks to manage political deepfakes.

• Mandatory Advertiser Disclosures: Most major advertising platforms now require political campaigns to explicitly disclose if an advertisement contains digitally altered or AI-generated imagery, audio, or video. Failure to do so typically results in the removal of the ad and potential account suspension.
• Content Labeling and Removal: Platforms have updated their terms of service to label suspected AI-generated content. While satirical or clearly marked synthetic content is often allowed to remain with a label, platforms actively remove deepfakes that violate voter suppression policies, such as fake audio messages giving incorrect polling locations or dates.
• Technical Provenance Standards: Tech companies are increasingly adopting standardized digital watermarking and metadata protocols, such as those defined by the Coalition for Content Provenance and Authenticity (C2PA), whose technical specifications were released in 2022. These standards embed data into a file at the moment of creation, allowing platforms to detect and label AI-generated media even if a user attempts to pass it off as authentic. The broader Content Authenticity Initiative (CAI), founded in 2019 by Adobe, The New York Times, and Twitter, promotes these standards with curbing disinformation as a core motivation.

Key Challenges in Enforcement

Despite the rapid rollout of these regulations, enforcing rules against political deepfakes remains highly complex.

• The Detection Arms Race: As AI models become more sophisticated, they generate content with fewer digital artifacts. Detection software must constantly evolve to identify new generation techniques, creating a persistent technological challenge.
• Jurisdictional Evasion: Deepfakes are frequently generated and hosted by bad actors or state-sponsored groups located outside the jurisdiction of the targeted election, making legal prosecution extremely difficult.
• Free Speech Protections: Drafting legislation that effectively bans deceptive political deepfakes without infringing on protected speech, such as political satire, parody, or legitimate artistic expression, requires precise legal language and often faces constitutional challenges in courts.
• Limits of Voluntary Commitments: Industry voluntary accords, while a meaningful step, are largely symbolic and are not a substitute for enforceable regulations and external oversight.

Summary

The regulation of AI deepfakes in elections is a rapidly evolving, multi-layered effort. It relies on a combination of government legislation mandating disclosures, technology platforms enforcing strict content policies, and the adoption of technical standards for digital provenance. While enforcement challenges remain significant, these emerging frameworks aim to provide voters with the transparency necessary to navigate an increasingly synthetic digital landscape.