MOSCOW, December 16. /TASS/. More than 80% of the largest financial companies will implement AI Security training programs in 2026, according to research conducted by the Fintech Association and Swordfish Security, available to TASS.
AI Security is the field protecting artificial intelligence systems from manipulations and supporting their reliable operation. It comprises protection of algorithms against hacker attacks, keeping educational data confidential, and preventing AI abuse.
"The research shows that fintech market participants plan to focus more on the rollout of AI Security in 2026. Specifically, 88% of respondents report that they are going to work on higher awareness and training of their staff in this sphere, 75% will deal with development of the AI Security internal policy, 63% will be involved in integration with protection tools, and 62% will pursue an audit or testing of AI systems," the research says.
According to the research data, 81% of Russian fintech companies identify machine learning and large language models as the core of their AI strategies. Meanwhile, more than a quarter of respondents have already experienced several security incidents related to attacks against AI systems.
The main driver of AI Security educational programs is the growing risk of confidential data leaks and difficulties in providing security when working with language models, the research indicates. "Rapid rollout of artificial intelligence prompts the financial sector, which is the most sensitive to potential attacks, to take prompt measures for retraining employees and implementing protective measures," the research notes.
"Artificial intelligence transforms the cybersecurity approach. The threat model changes, the surface of attacks becomes different, and therefore traditional strategies and approaches already require updating immediately. AI requires a new security logic, one combining principles of cybersecurity, risk management, and ethical regulation. It is impossible to stop the evolution of technologies, but processes can be created that are capable of making such development controllable, sustainable, and predictable," Alexander Tovstolip of the Fintech Association says.