MOSCOW, March 18 – The Russian Ministry of Digital Development has unveiled a comprehensive draft law on artificial intelligence, opening it for public discussion. The legislation, slated to take effect on September 1, 2027, introduces a framework that could significantly alter how AI is developed and used within the country.

A central provision grants individuals the right to challenge decisions made by government bodies and state-owned companies using AI systems through an out-of-court process. If an AI system causes harm, citizens will be entitled to compensation under existing civil law, establishing a direct mechanism for accountability in automated decision-making.

The bill mandates that AI developers eliminate discriminatory algorithms and block the generation of illegal content. Operators of AI systems are required to conduct safety tests and clearly inform users of system limitations. Service owners must act swiftly to prevent misuse.

Notably, the proposal introduces formal definitions for 'sovereign' and 'national' AI models. These are defined as models developed, trained, and refined exclusively on Russian territory by Russian citizens and legal entities, using datasets also compiled within the country.

Another key rule requires all AI-generated audio and visual content to carry a special watermark label. Major social networks will be obligated to verify this labeling and must either tag or remove unmarked synthetic content.

The ministry stated the law aims to establish clear rules for developers, businesses, and the state, intending to protect citizens from hidden manipulation and biased algorithms.

Source: RIA Novosti