DESILO Unveils Advanced Encryption Scheme for Enhanced Data Privacy in AI Applications

Innovative Encryption Scheme Sets New Standards for Data Privacy in AI Applications

In a significant development for data security in artificial intelligence, DESILO introduces its cutting-edge Gentry–Lee (GL) scheme at the FHE.org 2026 Conference held in Taipei. Co-authored by Yongwoo Lee, Chief Scientist at DESILO, and Craig Gentry, the figurehead behind Fully Homomorphic Encryption (FHE), this fifth-generation initiative marks an unprecedented leap in matrix multiplication performance, a crucial operation in many AI systems. With organizations increasingly reliant on AI to handle sensitive data, the GL scheme promises to reshape the landscape of encrypted data processing, pushing forward the concept of Private AI, which enables AI systems to operate on encrypted data without exposing personal information.

The GL scheme represents a paradigm shift in how matrix multiplication is approached within homomorphic encryption frameworks. Historically, FHE research has focused on demonstrating theoretical feasibility; however, the GL scheme innovates this by fundamentally restructuring the mathematical operations that underpin matrix multiplication. This advancement is vital for powering Large Language Models (LLMs) that are integral to modern AI, where data privacy and efficiency are paramount. As the demand for privacy-preserving solutions grows, particularly in heavily regulated sectors like finance and healthcare, the GL scheme is well-positioned to address these challenges, promoting safer data practices while maintaining the analytical strengths of AI systems.

Yongwoo Lee underscores the importance of enhancing matrix multiplication techniques for the practical application of Private AI. The implications of this technology extend beyond the realm of academic research, suggesting a pathway for enterprises to adopt AI with the assurance that sensitive data remains secure during processing. The full details of the GL scheme are available in the IACR ePrint archive, inviting further scrutiny and potential adaptation of this revolutionary framework by businesses facing the demands of a data-sensitive environment.

As the industry prepares for increased AI deployment, DESILO’s innovation aligns with the urgent need for effective data privacy solutions. The GL scheme not only showcases the technological capabilities of homomorphic encryption but also paves the way for broader acceptance and integration of Private AI in everyday business practices. This development resonates with organizations striving to implement AI solutions while adhering to stringent data protection regulations, illustrating the growing synergy between technology and compliance in the evolving landscape of data privacy.