Russian Tech Executive Calls for Strategic AI Alliances Amid Global Tensions

Sergei Bezbogov, a senior technology executive at Russia's VTB Bank, has outlined a pragmatic path for the nation's artificial intelligence development, emphasizing selective international cooperation as a necessity. Speaking at a major industry conference in Moscow, Bezbogov argued that achieving sovereign AI capabilities requires partnerships, primarily with China, in two critical areas: large language models and computing infrastructure.

He noted that while open-source LLMs from Chinese companies currently outperform domestic Russian versions, they introduce sovereignty risks. Using homegrown models from firms like Yandex and Sber mitigates some concerns. However, Bezbogov warned that total isolation is not the answer. 'We also need to use what is best in the world, otherwise we will start to lag behind,' he stated, suggesting training Russian models on anonymized data while seeking collaborative avenues.

The computing power gap presents a more stark challenge. Bezbogov pointed out that supercomputers operating in the United States vastly outstrip Russia's available capacity. To close this divide, he pointed to existing joint laboratories with Chinese partners and the import of Chinese-made graphics processors. 'We are already, in principle, beginning to assemble Russian servers on Russian territory with Chinese GPUs,' he revealed, framing such cooperation as essential for progress.

For financial institutions, the tension between powerful AI tools and strict data security is acute. Bezbogov explained VTB's solution: deploying LLMs like YandexGPT and GigaChat in an on-premise model, where the software runs exclusively within the bank's own secure infrastructure. This approach, he said, allows the use of advanced AI while maintaining control over sensitive client information.

Source: RIA Novosti