[The Police File]Another Deepfake Sex Crime Emerges on Campus, AI Threats Accelerate

It has been confirmed that deepfake sexual crime materials, in which female students' faces were superimposed onto nude photos, were shared once again in the university community known as the ivory tower of knowledge. The Telegram group chat, reportedly operating since 2020, had about 1,200 participants, and most of the victims were current students or alumni of Inha University. In the chatroom, personal information such as victims' contact details and student ID numbers were shared without filtering, and deepfake voice files featuring the victims' voices saying sexually charged words like "noye" (slave) or "juin-nim" (master) were also posted.

The existence of the chatroom became public when a victim, an Inha University graduate, began receiving repeated messages since the end of last year asking, "I saw you in the chatroom; is that really you?" The victim immediately filed a complaint with the Michuhol Police Station in Incheon, and in February of this year, the university requested an investigation from the Incheon Police Agency. However, the main perpetrators, including the chatroom operators, have yet to be identified. This is because the Telegram servers are overseas, making it difficult to pinpoint individuals.

Ultimately, the victim entered the chatroom themselves to collect evidence and submitted it to the police. One man who downloaded and distributed the victim's deepfake photos was prosecuted for violating the Sexual Violence Punishment Act and sentenced to one year in prison, but that was all. Another Telegram participant was additionally apprehended by the police but was not prosecuted. This is because, for images, videos, or audio involving a person's face, body, or voice, clear intent to distribute is required for punishment. The production or viewing of deepfake pornography for personal possession is excluded from punishment.

In fact, there is no concrete solution to counter the malicious use of deepfake technology. The "AI Industry Promotion and Trust-Based Foundation Act (AI Basic Act)" failed to pass in the 21st National Assembly. Six bills have been proposed in the 22nd National Assembly but remain pending. An amendment to the Information and Communications Network Act, which would mandate watermarking when posting AI-generated content such as deepfake videos or audio online, also seems far from realization.

This contrasts sharply with major countries worldwide actively revising regulations related to AI. In March, the European Parliament finally approved a fundamental law that promotes AI technological innovation while ensuring safety and respect for fundamental rights. The United States also saw President Joe Biden sign an executive order last October on the development and use of AI with safety, security, and reliability. Neighboring Japan discussed the need for generative AI regulation at the G7 summit last year, and China is also enacting laws concerning generative AI services.

Deepfake videos are a serious social issue that can lead not only to digital sexual crimes but also to violations of portrait rights, personality rights, privacy, and defamation. The Korea Communications Standards Commission detected over 6,000 cases of "sexually false videos" misusing deepfake technology in the first half of this year alone, accounting for 84% of last year's total. On the other hand, among 71 first- and second-instance rulings related to deepfake crimes from 2020 to 2023, only four resulted in prison sentences, indicating minimal punishment.

Considering the pace of technological advancement, deepfakes are likely to become more sophisticated and widespread. This means the number of digital sexual crime victims could increase exponentially. It is time to implement technical, judicial, and social regulations to address deepfakes before it is too late.