South Korea’s AI Basic Law, which went into effect at the start of this year, is already facing its first major revision attempt. While the original legislation requires tech providers to notify users when an image, video, or audio clip is computer-generated, the rules have proven inadequate in practice. Lawmakers have identified a legal loophole they are openly calling a “blind spot” in transparency regulations.
Current regulations don’t specify where or how the disclosure of a material’s synthetic origins should appear. As a result, most tech companies just add small captions or icons strictly within their own app interfaces. The problem hits when secondary creators screenshot, crop, or download a video to their drive, stripping away the informational label entirely. The existing law hasn’t treated these external actions by internet users as a clear-cut violation.
To shut this practice down, Representative Kim Dai-sik, along with nine other lawmakers from the opposition People Power Party, has introduced a new legislative bill. The document radically shifts the approach to content tagging. The focus moves from visual platform notifications to an absolute mandate to embed identifiers – defined as specific codes, letters, or symbols – directly into the generated file’s code. The watermark is meant to serve as an integral part of the media, sticking with it regardless of any subsequent sharing.
Under the proposed bill, removing, damaging, forging, or altering a mandatory watermark in any way will be treated as a criminal offense. These actions carry penalties ranging from fines of up to 20 million won (roughly $13,500) to as much as two years behind bars.