iTPROTECT Launches Suite of Tools to Secure AI Systems
The portfolio includes 15 products across five key areas.
More Russian companies are integrating artificial intelligence into their IT infrastructure. While AI can significantly streamline work, experts warn that without proper safeguards, these tools can become a serious source of information security risks. As a result, the market is seeing a growing range of tools designed to reduce the risks associated with AI use.
Blocking Potentially Dangerous Prompts
iTPROTECT, a Russian IT integrator, is now entering this market. The company has begun offering AI security solutions, with a portfolio that currently includes 15 products grouped into five key areas.
AI Firewall is designed to monitor interactions with internal and external AI systems. To prevent data leaks, it can automatically block unauthorized prompts and responses. The solution detects and masks sensitive data, restricts employee access to it, and filters language model outputs.
The portfolio also includes AI/ML automated penetration testing, which assesses AI models and applications against 40 types of attacks. Applications can be tested before they go live. The tool also includes a dedicated feature for scanning for potential data leakage threats.
Monitoring AI Agent Activity
Specialized AI agent security tools let companies monitor agents in real time, taking their privileges and security policies into account. They track not only what agents do, but also which information sources they use. The tools can block dangerous commands and detect anomalous prompts. This gives corporate security teams visibility into which AI tools are running across their infrastructure and helps reduce the attack surface.
As part of the new offering, the iTPROTECT team also deploys secure enterprise AI agents within a customer’s protected environment using a specialized platform. This approach helps address “shadow AI,” in which employees use external AI tools without oversight from their company’s security team.
Another area focuses on large language model (LLM) testing. The tests assess models’ resilience to jailbreaks, multistep attacks and attempts to bypass security controls. They also examine whether confidential information can be extracted through documents and prompts, and assess the risks of unauthorized data access, tool misuse and privilege escalation.
Responding to Market Changes
Company representatives say the new offering reflects changes in Russia’s IT market. According to an iTPROTECT survey, nearly one in four Russian companies (24%) already uses AI actively in its workflows or considers adoption a priority. Among this group, 40% of information security leaders believe AI expands the attack surface and are developing measures to protect their businesses against the resulting risks.
In Russia, the work to address security risks associated with AI has been for conducted several years. In 2024, Kaspersky Lab announced a training course on the safe use of neural networks through its Kaspersky Automated Security Awareness Platform. Designed for employees, the program required no specialized technical knowledge.
In June 2026, the Bank of Russia published guidance on information security in the development and use of AI in the financial sector. The document organizes the risks associated with AI and recommends that financial institutions develop their own threat models and information security policies. For high-risk, mission-critical operations, it recommends leaving all key decisions to human employees rather than delegating them to automated systems.
Experts expect demand for AI security in Russia to grow as these tools become more widely adopted. Developers are already moving beyond standalone products to offer entire ecosystems designed to make AI deployment as secure and predictable as possible.
