Anton Redozubov: “Moscow has become a testing ground for the AI revolution”

— What were the key AI market highlights of the past year?

— 2025 marked a turning point for Russia: we stopped looking at AI as at some sort of a foreign gimmick and adopted it as a critical technology of national scale. If we highlight three main results of the year, then the first one will be the trend towards sovereignty. Major players have finally placed their bets on their proprietary developments, designed for Russian language and domestic context.

The second result is the migration of risks: AI made routine tasks, like speech recognition or translation from foreign languages, easier, but at the same time gave rise to new threats — deep fakes, sophisticated fraud, infrastructure attacks. And the third result is the transtion from theory to practice: now there are first domestic benchmarks and industry solutions. Sber and Yandex presented their speech-to-text models, business got efficient work tools.

— Has AI mitigated any risks?

— It did not, but it transformed risks, making them more complex and requiring new competencies.

— It sounds like an encouragement to learn rather than to be afraid. But aren’t we overly optimistic?

— We certainly shouldn't go into seclusion and become Luddites — let’s recall English workers of the XIX century, destroying machinery, “denying” them their work, or means of subsistence. History doesn't repeat, but it can rhyme if we misuse the IT tools at our disposal. Our goal is not to reject it, but to learn to control it. AI is not a magic wand, it is a lever. If we use it the right way, it will multiply our force, and if we don’t — it will hit us hard.

— Late last year Danish Saxo Bank predicted introduction of a “virtual CEO” at the helm of a major corporation in 2026. Could this be real?

— In terms of technology capability – yes, it could, it is already possible to implement such a solution today. But in terms of legal and executive concerns it is definitely impossible, especially in Russia. AI cannot be held responsible for its decisions, and this is the cornerstone for any governance and legal regulations. In my opinion, reality will be more mundane: instead of a “virtual CEO”, we will see advanced decision making support systems, which currently run up to 80% of analytical  operations. The key issue here is control. “AI Cleanup Assistants” will be required to audit algorhytmic solutions and errors. And even if a position of Vibe Code Cleanup Specialist — a person cleaning up AI-generated code errors (some analysts say this is one of the “jobs of the future”) is still a meme, but it is terrifyingly close to reality. 2026 will be the year of experiments in this area, but there will be no large-scale transition: legal and reputation risks both for major corporations and government agencies are too high.

— Will AI replace government officials, at least partially?

— It won’t definitely happen in 2026. AI can and must serve as a tool for application preprocessing, data validation, anomaly detection. And it has been doing this for a long time, every day, without any hype. Decision making support systems were developed way back in 1980s. Yet, final authority, especially in socially sensitive areas, such as approval of social benefits, issue of permits, should stay in human hands. This is insurance against data or algorithm logic errors. Technology helps, but it doesn't replace responsibility.

— What, in your opinion, is the main 2025 takeaway for business and government?

— The last year has shown that those who started systemic implementation of the technology, are already reaping benefits, and those who waited for a “perfect time”, are lagging behind. But it is important to remember: AI sovereignty is not isolation, but the ability to develop one's own solutions based on global experience, while taking national interests into account. The risks haven't gone away, they've just become more complex. And this means that we need more than just engineers; we need lawyers, ethicists, and cyber security professionals who understand how 'black-box' systems function. 2025 gave us a map, and 2026 will require us to navigate this map.

— Why do some still see artificial intelligence as a threat, while others see it as a miracle?

— When I’m asked about AI, I often recall a quote by William Gibson: “The future is already here — it's just not very evenly distributed”. Some see a neural network as an image generator, others view it as an unobtrusive assistant sorting mail or plotting a route. I'm being pragmatic about this: AI is neither a miracle nor a monster — it is a tool, just like a steam engine of old or a personal computer. It’s only a matter of adoption speed: previously the humankind had enough time to get used to a new technology, and now everything happens in a blink of an eye. Articifial intelligence in Russia is everyday reality, it operates in support desks, navigators and security systems. The question is not whether AI will exist, but how deeply it will be integrated into our daily routines.

— Many people are concerned that algorithms will take away their jobs. How valid are these concerns?

— AI will replace non-professionals, those who resist development, and not professions. This evokes a historical parallel: accountants used to work with abacuses, then calculators, followed by the era of Excel and 1C. The profession has not disappeared, it has evolved. Any routine white-collar jobs will be under threat: template-based copywriting, basic accounting, data input operators. Those who use AI as a personal assistant, have nothing to fear over the next five years. A doctor, monitoring diseases with the help of algorithms, or an engineer, designing a part in dialogue with a neural network, will only be more efficient. As for a full vehicle autopilot, we are still a long way away from infrastructure that would enable it, where it would be contolled by “smart road”, and not just the car.

— Can Russia become a global leader in neural networks, considering the sanctions on purchasing foreign technologies and other obstacles?

— Sanctions are a challenge, historically answered by our people with a major breakthrough effort. Football coaches say: “The Russians only start playing after conceding the first goal”. Our strengths involve profound mathematical school and unique data: computer vision for industrial applications, geoanalysis, agricultural technologies and voice technologies. Cybersecurity will play a special role, it will require AI solutions for protection and identifications of fraudsters. It is crucial to have affordable technologies. In this case major business operates as a nuclear icebreaker: it clears the way and invests in fundamental models. But it would be small business, which will deliver solutions to every person like a swift boat. Any attempt to “bottle up” AI within corporations will strangle the market.

— You are a security expert. What are the key risks posed by AI expansion and how do we protect ourselves from these?

— If AI could help in search for medicines, it would be as likely to help with fraud. I mean it is not the tool itself, but those who use it and for what purposes. This is the eternal battle of sword and shield. The key “dark side” of current IT solutions are deepfakes and fraud. An average person doesn't need to be a cybersecurity expert to protect themselves from all this: practicing digital hygiene is enough. First of all, one should develop critical thinking: If a "relative" asks for an urgent money transfer, call them back on their known, existing phone number to verify the situation. Second, establish a code word with your family for emergencies as a reliable defense against voice scams. Third, maintain basic tech habits: use two-factor authentication procedures and create strong passwords to access significant personal information.

— What should we do with education? How do we prepare our children for the world, where answers are provied by a machine?

— The most important issue would be to teach a child to think and ask the right questions. AI provides answers, but only a human can pose a task. Philosophy, logic, mathematics and natural sciences are fundamental for this purpose. Hard skills matter, but you need an understanding of how algorithms work, not just programming. And apart from these, soft skills are critical too: communication, empathy, creativity. And naturally, attutude towards gadgets: a phone has to be used as a tool, not as a substitute for reality. Live communication, sports, and development of fine motor skills are key to develop a healthy neural network in your head and a sound mind in a sound body.

— Moscow is actively implementing AI in governance today. How successful is this experience?

— The capital city is the main testing ground and incubator for the latest IT practices. Moscow’s comprehensive approach makes it unique: the city is building a complete digital ecosystem, which is becoming a model for Russian regions. The city sets world standards in transport: 2026 will see the start of a pilot project for integrated traffic control in Tushino, the share of “smart” traffic lights will increase up to 80%, and biometric payments will be available for all ground transport modes.

Digital twins are used in construction to verify compliance with standards and codes before construction starts.

In healthcare, AI in the EMIAS (Unified Medical Information and Analytical System) is already analyzing X-ray images, helping to identify pathologies in early stages and relieving doctors of routine work.

A special experiment in the field of state supervision is scheduled  for 2026–2028: neural networks will automatically identify landscaping and environmental violations through CCTV and drone cameras. This will eliminate the human factor and corruption. Moscow operates not as a “city under a golden dome”, but as a federal competence center scaling successful solutions nationwide.

— What does this mean?

— Successful practices are already yielding measurable results. The Moscow Region, following the lead of Moscow, saved over ₽2,5 billion 2025 under 67 AI-driven projects. Neural networks have been insstrumental in identifying over 500 thousand property maintenance violations, and by the end of 2026 the number of such projects in the Moscow Region will grow up to 80. As highlighted by Andrey Vorobyov, Governor of the Moscow Region, the goal is to “free people from routine”. The “Gorodovoy” Complex operates in a similar fashion in Saint-Petersburg: this winter it monitored 21 types of violations, ranging from ice cover to failure to remove snow in a timely manner.

— How systematically will this experience be scaled up at the federal level?

— The process is already underway. Oleg Morozov, Chairman of the State Duma Committee on Control, made a statement: the outcomes of the Moscow neural network experiment will be reviewed by state supervision agencies, and the “most successful practices” will be rolled out in “other regions”. Now the federal authorities are preparing to introduce KPIs for AI use for all Russian regions, and the “Digital Region” platform of the Moscow Region is viewed as a hub to exchange best practices.

The instructions from the President of the Russian Federation dated January 3, 2026, following the "Journey into the World of Artificial Intelligence" (AI Journey) conference, actually take this subject from the status of “interesting pilot projects” to the mode of state oversight of performance outcomes: there will be a uniform implementation database, and it is planned to include an AI parameter into the digital transormation ranking of the Russian regions. Simply put, what works in Moscow and the Moscow Region today will become the national standard tomorrow.

— What is your personal outlook for the next year: will there be “digital paradise” for us?

— We all believe a little bit in a magic wand that will turn us from Ilya Oblomov into Ilya Muromets. But the reality offers two scenarios. The pessimistic scenario, let’s call it “Digital Swamp”: we will stop straining our brains by entrusting everything to algorithms. This is a direct path to cognitive degradation and increased unemployment for those who failed to retrain for a new profession. In a word, just like the plot of the Idiocracy movie. But there is another option, an optimistic scenario: AI will become a tool which will liberate us from routine tasks. The labor market is already adapting to the new IT reality, and given the current labor shortages in the economy, people will be better off if they invest into retraining, rather than remain unemployed.

— What will be the outcome?

— We already have some of the “digital paradise” built around us — in a couple of clicks we can order food delivery or a tour to the Baikal Lake. The main challenge of 2026 lies not in technology, but within us. AI will not replace humans, but humans, using AI, will replace those who reject it. Real progress will arrive when we quit waiting for miracles and start building it by ourselves, combining perseverance, self-confidence and critical thinking. This is that very “humanity”, which cannot be replicated by AI.

* Expert estimates show that by 2030 the Russian AI market could grow up to ₽778 billion, registering average annual growth of 68,1%.