Install a neural network: how AI will help solve crimes in Russia
By 2030, 90% of crimes in cities will be solved using artificial intelligence (AI) technologies, experts have predicted. Video analytics using neural networks alone can prevent violations of the law, while the overall potential of AI technologies to improve security is very high. For more information about how artificial intelligence will help solve crimes in Russia and what its prospects are in law enforcement work, see the Izvestia article.
What is interesting about AI for fighting crime?
Today, the global biometric services market is showing steady growth, and one of the most sought-after areas of AI video analytics development is naturally the security sector, says Alexey Palamarchuk, CEO of NtechLab, in an interview with Izvestia.
"Many cities and governments, both Russian and foreign, see in AI technologies an extremely high potential for improving security, since the very fact that video analytics are available in the city prevents crimes," the expert notes.
A person, knowing that AI is able to find it even in a crowd, on a large square, and "see" it on the transport infrastructure, will not even dare to break the law — for example, to steal something or steal a car, Alexey Palamarchuk notes. According to the expert's forecast, by 2030, 90% of crimes in cities will be solved using AI technologies.
Artificial intelligence in terms of combating offenses is interesting for its ability to instantly analyze huge amounts of data, replacing tens of thousands of people, adds Nikita Leokumovich, head of the Department of Digital Forensics and Cyber Intelligence at Angara MTDR. In particular, AI can help identify hidden patterns, predict likely crime locations and times based on historical statistics, and recognize faces, gait, voice, and even deepfakes in real time.
How neural networks are already helping law enforcement agencies
Today, AI-based technologies are already helping law enforcement agencies both in Russia and in other countries of the world. According to Alexey Korobchenko, head of the Information Security Department at the Security Code Company, in our country we are primarily talking about online video analytics.
— Similar technologies are used in Russian cities — they allow you to monitor security on the streets, record violations and prevent possible crimes, for example, by noting the presence of unattended items, — says the interlocutor of Izvestia.
Today, Russia is actively developing digital footprint analysis systems to combat cybercrime, adds Irina Mezheneva, a leading analyst engineer at Gazinformservice. In particular, we are talking about monitoring the darknet, automatic search for phishing resources and systematic analysis of suspicious financial transactions in the banking sector. These technologies have already become a de facto part of the daily work of law enforcement agencies, the expert notes.
According to Alexey Korobchenko, in terms of the use of AI technologies in law enforcement, Russia is keeping pace with other advanced countries in the world, where the main focus is also on video analytics. This practice is especially common in the USA and China.
"In these countries, smart cameras make it possible to detect criminals, record cases of firearms and other weapons being used, analyze data and map crime—prone areas for further development of counteraction measures," the expert says.
The United States is already implementing AI—based risk modeling systems - in other words, predicting violent crimes based on previously identified statistical patterns, Nikita Leokumovich notes. Such a system has already reduced violent crimes with guns by 22%.
What are the pros and cons of AI policing
According to experts interviewed by Izvestia, the main advantage of using neural networks by law enforcement agencies is a reduction in the number of crimes. According to Alexey Korobchenko, in Moscow alone, the number of car thefts and thefts decreased by 97% in 13 years (from 2012 to 2025). This did not happen without the help of AI technology systems.
"The advantages of this practice include a significant increase in the speed and accuracy of data analysis, the ability to prevent crimes before they are committed, and optimization of police resources," Nikita Leokumovich lists.
Irina Mezheneva agrees that the use of AI technologies in the work of law enforcement agencies has great advantages. She pays attention to the speed and objectivity in processing facts: the machine does not get tired and has no personal animosity. At the same time, the disadvantage that overrides many advantages is the catastrophically high cost of error. If the AI makes a mistake in recommending a movie, the user loses a couple of minutes, but if it incorrectly identifies the perpetrator, human freedom and life are at stake, the expert emphasizes.
At the same time, according to the editorial interlocutor, in the United States and Europe, the practice of "predictive Policing," when algorithms predicted the likelihood of crimes in certain areas, faced harsh criticism due to algorithmic bias. It turned out that models can "inherit" human biases, which led to discrimination.
"The technology's success here is relative: it works great for finding missing people, but it remains controversial in matters of preventive control,— the expert warns.
In turn, Alexey Korobchenko draws attention to the fact that the main disadvantage of AI technologies in the service of the law enforcement system lies in the area of total control, which is most clearly manifested in the PRC's policy against residents of the Xinjiang Uygur Autonomous Region. International reports by human rights organizations say that the region is a "digital concentration camp" that operates primarily thanks to AI cameras.
What are the prospects for law enforcement AI technologies
Automation of evidence collection and database search using AI technologies in law enforcement agencies will indeed become more widespread in the coming years, but crime detection is a complex legal and operational process that does not consist solely of identification by cameras, Irina Mezheneva says in an interview with Izvestia.
"The prospects for AI in Russia will depend not so much on the perfection of algorithms as on the creation of a transparent legal framework and control mechanisms," she believes. — The main challenge is not to teach a neural network to find faces, but to create a system where every AI decision will be subject to mandatory human validation, eliminating the risk of an automated judicial error.
Meanwhile, according to the expert, today modern neural network models remain "black boxes", when working with which the result is visible, but the logic of the conclusion is not always transparently traced. Without the introduction of explicable AI (XAI) methods and a rigid human verification loop, there is a risk of creating a system that will make verdicts based on statistical errors rather than real evidence, the expert warns.
Nikita Leokumovich, in turn, points out that AI technologies require appropriate related infrastructure for successful operation, for example, video surveillance cameras, stable network connectivity and the equipment on which the systems will operate. However, in the current reality, it is often very difficult to ensure all this.
— Finally, it is necessary to change the metrics for assessing the success of law enforcement agencies, because at the moment they take into account the number of crimes solved, not prevented, — concludes the interlocutor of Izvestia. — Therefore, it is possible that their interest in the implementation of such systems will remain at a low level, as it will "worsen" the main indicators of their work.
Переведено сервисом «Яндекс Переводчик»
