MugendAI’s new QEV implementation service uses pre-transmission detection and controls to prevent information leaks when companies use generative AI and AI agents. The service will launch on 1 November 2026, leveraging the QEV model’s high detection performance.
MugendAI LLC announced that it will begin offering its QEV implementation service on 1 November 2026 as a measure to prevent information leaks when companies use generative AI and AI agents. The service automatically checks whether data being sent outside the company contains personal or confidential information and prevents leaks before transmission.
The QEV implementation service is centered on the “QEV” decision-making AI model developed by QueryPie AI. MugendAI will participate as a sales partner supporting customers from implementation through operation. The QEV model is said to have a high detection rate for documents containing confidential information, with detection performance at an extremely high level. It is designed with consideration for Japan’s personal data protection framework and corporate information management practices.
As the use of generative AI expands, concerns have been raised that the risk of information leaks is increasing, highlighting the importance of screening data transmissions between internal and external AI systems. The difficulty of determining how data sent to external AI services is stored has been identified as a particular risk factor.
The QEV implementation service enables pre-transmission screening between internal systems and external AI systems, and designs and implements decisions and transmission controls based on the definitions of personal and confidential information. In an internal evaluation by QueryPie AI, the detection rate for documents containing confidential information reached 99.69%, indicating significantly higher effectiveness than external models in terms of missed detections.