US Calls South Korea's Deepfake Law 'Censorship' as AI Rules Fracture

The US State Department publicly criticised South Korea's deepfake criminalisation law on Tuesday, calling it "a form of censorship that chills free expression," in a statement that exposed a widening rift among democratic allies over how to regulate AI-generated content.

South Korea's law, enacted in January 2026, makes it a criminal offence to create or distribute sexually explicit deepfakes without consent, carrying penalties of up to seven years in prison. The legislation followed a nationwide crisis in 2024 when AI-generated explicit images of Korean women and girls circulated on Telegram channels, prompting mass protests.

"We respect Korea's sovereign legislative process, but laws that criminalise synthetic media creation risk suppressing legitimate speech, satire, and artistic expression," said a State Department spokesperson in a written statement provided to Reuters.

Allied Approaches Diverge

South Korea is not alone. The European Union's AI Act, which took effect in phases through 2025, requires labelling of deepfake content and imposes fines for non-compliance. The United Kingdom's Online Safety Act criminalises sharing deepfake intimate images. Australia passed similar legislation in late 2025.

The United States has taken the opposite path. The Trump administration dismantled the State Department's Global Engagement Centre, which had coordinated counter-disinformation efforts, as part of a broader rollback of government involvement in online content moderation.

A settlement reached in March with social media companies that had partnered with government counter-disinformation programmes effectively ended federal involvement in flagging synthetic media. The administration launched Freedom.gov, a portal for reporting government censorship, as a replacement.

"The US position is now that deepfakes are a speech problem, not a technology problem," said Renée DiResta, a researcher at the Stanford Internet Observatory. "That puts Washington at odds with virtually every other democracy."

Korea Responds

South Korea's Ministry of Science and ICT issued a measured response, noting that the law was drafted in consultation with constitutional scholars and targeted non-consensual intimate imagery specifically, not political speech or satire.

"Our law protects victims. It does not restrict political expression," said the ministry in a statement. "We are disappointed that an ally would characterise victim protection as censorship."

Korean media coverage was sharp. The Chosun Ilbo, Korea's largest-circulation newspaper, ran an editorial headlined: "America Lectures Us on Deepfakes While Its Own Women Are Unprotected." The piece noted that the United States has no federal deepfake law and that fewer than half of US states have passed legislation addressing AI-generated intimate images.

The Korea Herald reported that the US criticism had become a trending topic on Korean social media platforms, with many users pointing out that the original deepfake crisis disproportionately targeted Korean school-age girls.

Regulatory Fragmentation

The disagreement matters beyond bilateral relations. Without a shared framework among democracies, technology companies face contradictory legal requirements across markets. A deepfake video legal in the United States may be criminal in South Korea and subject to mandatory labelling in Europe.

"Companies will comply with the strictest jurisdiction they operate in, or they'll fragment their products by market," said Matt Sheehan, a fellow at the Carnegie Endowment for International Peace. "Either way, the user experience of AI content will depend on where you live."

China has its own deepfake regulations, implemented in 2023, which require watermarking of all AI-generated content and government approval for certain synthetic media applications. Russia has no comprehensive law but uses existing information security statutes to prosecute deepfakes selectively.

Counter-Disinformation Vacuum

The timing of the US criticism coincided with the dismantling of its own counter-disinformation capacity. The Global Engagement Centre closure left no federal entity responsible for tracking foreign deepfake operations, according to former GEC officials who spoke to the Washington Post.

This gap is acute during the Iran conflict. Iranian state media has deployed AI-generated imagery in its war communications, according to the Atlantic Council's Digital Forensic Research Lab. Without the GEC, no US government body is systematically tracking or countering these operations.

"We are asking allies to stop regulating deepfakes while simultaneously losing our ability to detect them," said former GEC acting director James Rubin in testimony to the Senate Foreign Relations Committee last week.

The next scheduled US-Korea bilateral technology dialogue is set for May. South Korean officials said the deepfake law would not be on the agenda for renegotiation.