Russia Launches a New Service That Anonymizes Documents Before They Are Sent to AI Models
Teams from CorpSoft24 and Corp Lab have developed a service that “hides” personal and sensitive data in documents before they are processed by neural networks. The approach allows employees to use tools such as ChatGPT or YandexGPT without exposing personal data or trade secrets.
A Response to Growing Market Demand
The solution enables the safe use of external AI tools, including ChatGPT and YandexGPT, by preventing the transfer of personal data and confidential business information. The service automatically removes sensitive content from documents, reducing the risk of data leaks during analytics workflows and document processing.
The product addresses a rapidly growing demand for data protection as AI becomes embedded in corporate processes. It acts as a buffer between confidential information and cloud-based AI services, a capability that is increasingly critical for information security in highly digitized environments.
Toward an Industry Standard
Although the prototype is primarily aimed at the domestic corporate market, it has notable export potential. In theory, the technology could be relevant for jurisdictions with strict data protection regulations comparable to GDPR, where data privacy is a key constraint on the use of AI services. Reaching those markets, however, would require significant preparatory work. The developers are also considering expansion into CIS countries, where demand for secure AI tools in enterprise environments is growing, as well as interest from international companies facing similar data leakage risks.
Within Russia, the technology aligns with enterprise needs around digital transformation and the regulated use of AI. It could become a standard solution for organizations with elevated data protection requirements, including banks, law firms, and government agencies. Future plans include integrating the anonymization technology into existing enterprise content management (ECM), business process management (BPM), and data loss prevention (DLP) systems. At present, an MVP prototype has been built and is undergoing internal testing and refinement, with plans for deployment on dedicated servers with centralized access.
The Direction of the Sector
Related approaches to data anonymization and secure AI use include automated anonymization based on machine learning. In these systems, machine learning techniques automatically identify and remove sensitive information to reduce the risk of data leakage. Methods include automated detection of confidential data, context-aware anonymization, and synthetic data generation. Such systems are already used in large financial institutions.
Data confidentiality in AI systems is also being addressed through research into privacy-preserving machine learning. These efforts rely on techniques such as differential privacy for text, which limit a model’s ability to expose original source data while preserving functionality. At the core of this approach is differential privacy as a formal framework.
The DeepPrivacy technology, for example, uses generative adversarial networks (GANs) to replace faces in images with synthetic ones. While it does not yet have a direct commercial implementation, it serves as an important reference point for how automated personal data protection technologies may evolve.
Traditional DLP products focus on protecting data within organizations by preventing leaks through external communication channels. While they share objectives with AI-driven de-identification tools, DLP systems represent a broader class of conventional security solutions and typically do not emphasize the use of artificial intelligence. At the same time, tightening AI security requirements in Russian and European jurisdictions are shaping the rules for data protection in this area.
A Foundational Enterprise Technology
The document anonymization service developed by CorpSoft24 represents a meaningful step forward for enterprise security in the context of AI adoption. It addresses a critical challenge – protecting personal and commercial data from leaks when working with external AI services – and enables safer integration of artificial intelligence into business processes.
The developers expect the product to become a foundational enterprise technology for secure AI usage, with growing demand anticipated from the banking sector, government bodies, and legal firms. In international markets, the service could complement existing solutions for secure AI adoption. At the same time, it will need to compete with established vendors and meet regulatory requirements such as GDPR and Russia’s Federal Law No. 152 on personal data.
