Back to feed

Putin's military AI push masks Russia's lag in frontier model race

2 min
Putin's military AI push masks Russia's lag in frontier model race

This digest was compiled by AI from multiple sources — links to the originals are below.

Russia is deploying AI for censorship, drone warfare, and domestic surveillance, but lags far behind the US and China in developing cutting-edge models. President Vladimir Putin touts sovereign AI capabilities, yet Russian models like Sberbank's Gigachat trail US systems by a year. The Kremlin's focus on applied AI benefits the state while frontier research remains uncompetitive.

Key Facts

  • Eksmo-AST, a major Russian publishing group, trained an AI agent to screen 20,000 books for censorship risks after Putin's 2022 invasion of Ukraine.
  • Sberbank CEO Herman Gref said the bank's Gigachat AI model is about six months behind China and a year behind the most advanced US models.
  • Russia has at least two major large-language models: Alice from Yandex and Gigachat from Sberbank.
  • Putin claimed Russia is one of only three countries actively developing sovereign AI capabilities.

AI for Censorship

After Putin imposed harsh censorship laws following the 2022 invasion of Ukraine, Eksmo-AST trained an AI agent to identify books and passages that risked violating the new restrictions. The publisher conducted approximately 20,000 checks, and 600 volumes were referred to external experts for additional review, according to group president Oleg Novikov. Novikov told Russian state television that the company is 'on the frontier of technological progress' in applying AI to censorship.

Military and Surveillance Applications

AI-powered drones are increasingly used by Russia to bomb Ukraine, and bots create fake social-media profiles to spread the Kremlin narrative abroad. At home, facial-recognition surveillance networks track Russian citizens, serving as a source of pride for Putin, who frequently mentions AI in speeches. Russia is also developing applied AI for state services, industrial production, and healthcare, including individual cancer vaccines similar to efforts in the US and UK.

Lag in Frontier Models

Russia's two main large-language models are Alice, developed by Yandex, and Gigachat from Sberbank, the country's largest lender. Sberbank CEO Herman Gref, a long-time Putin ally, said the bank's AI is about six months behind China and a year behind the most advanced US models. The clearest sign that these models are not competitive is that hardly anyone outside Russia uses them.

1 source

Time · lag behind first