⚖️ Mandatory AI Content Labeling Proposed in Russia
Russia is considering stricter regulations for generative artificial intelligence. Labeling requirements for images, audio, and video created using large-scale AI models may shift from a previously proposed voluntary format into a strict legal mandate.
A relevant draft amendment to the Law on AI and the Civil Code was prepared by a working group under the Presidential Council for Codification, led by Pavel Krasheninnikov.
Under the proposed legislation, the phrase “created with the use of” could apply even to media only lightly edited by a neural network, as the draft does not define a minimum threshold for AI involvement. Furthermore, the exact format, enforcement procedures, and responsible regulatory body have yet to be designated, leaving open questions regarding liability for sharing third-party generated content and the preservation of metadata during cross-platform file transfers.
The current version of the framework Law on AI—where the labeling provision is set to take effect on March 1, 2027—only requires major social networks and online services with over 500,000 daily users to provide the technical capability to apply labels.
Industry representatives and tech experts have responded to the new initiatives with concern:
- Risk of driving activity into “grey markets.” Participants in the discussions note that a mandatory rule would burden domestic developers while failing to combat deepfakes, as malicious actors will simply continue using unlabeled foreign services.
- Technological immaturity. Karen Kazaryan, Director of the Internet Research Institute, pointed out that the global tech community is still searching for reliable solutions to keep machine-readable metadata intact when content is edited.
- Excessive bureaucracy. Major IT companies warn that even routine photo filters could fall under the new definition, while Alexey Byrdin, CEO of the Internet Video Association, highlighted the risks of legal uncertainty if core requirements are relegated to secondary regulations issued by the Ministry of Digital Development.
As an alternative, experts suggest looking to international precedents: China’s hybrid labeling system (which combines visible and invisible watermarks) or the provisions of the European AI Act, which require disclosing the synthetic origin of generated material upon publication.
Source: Vedomosti
