디사일로

DESILO has been honored with the “2026 H1 Innovation IT Award.” Through its fully homomorphic encryption (FHE)–based Private AI technology, DESILO has achieved the world’s first End-to-End Encrypted LLM, operating entirely on encrypted data from input to inference. This breakthrough fundamentally overcomes the limitations of existing approaches and redefines the benchmark for next-generation enterprise AI infrastructure—delivering both security and performance. DESILO will continue to lead the global standard for Private AI. 디사일로가 "2026 상반기 혁신 IT 대상"을 수상했습니다. 디사일로는 완전동형암호(FHE) 기반 Private AI 기술을 통해, 데이터 입력부터 추론까지 전 구간이 암호화된 세계 최초 End-to-End 암호화 LLM을 구현했습니다. 이는 기존 기술의 한계를 근본적으로 돌파한 성과로, 보안과 성능을 동시에 충족하는 차세대 엔터프라이즈 AI 인프라의 기준을 새롭게 정의합니다. 디사일로는 Private AI의 글로벌 표준을 선도하겠습니다. https://lnkd.in/g84ieYfF

Standardization is the key to enterprise adoption. Based on an interview with our own Ilaria Chillotti and Damien Ligier, our new article breaks down the state of Fully Homomorphic Encryption (FHE) standardization at ISO and NIST. Understand the critical business impact, from ensuring interoperability across vendors to building regulatory confidence for deployments.

We are happy to announce that the registration for the 9th Homomorphic Encryption Standards Meeting is now open. Click https://luma.com/r67depdk to register. The 9th HomomorphicEncryption.org Standards Meeting will be held in Seoul, South Korea on March 5-6, 2026. This meeting will take place just before the FHE.org 2026 (March 8) and RWC 2026 (March 9-11) conferences, which will both be held in Taipei, Taiwan. You can find the latest information about the upcoming #FHE Standards Meeting at https://lnkd.in/e-HWUwES. We are grateful to the local organizers from LG Electronics and DESILO Inc., including Min-Wook Jeong, Ilaria Chillotti, Damien Ligier, Myoungbee Sung, and Jun Hong Park. The event will include a keynote speech by Craig Gentry on his latest FHE advances. Looking forward to seeing many of you in Seoul! Rachel Player, Shai Halevi, Ro Cammarota, Kurt R Rohloff, Jung Hee Cheon, Kristin Lauter #homomorphic #privacy

The DESILO FHE Library v1.8.0 is now live, bringing a major speed-up in encrypted matrix computation. Highlights in this release: 📍GL (Gentry–Lee) scheme support for matrix-friendly operations and up to 80× faster encrypted matrix multiplication 📍Faster lossy/sign bootstrapping for real-number data These improvements make it easier to run matrix-heavy ML inference with homomorphic encryption from start to finish with a significantly improved performance. We’re also opening a community Discord channel for developers and researchers working on FHE, encrypted ML inference, and secure computation pipelines. Our engineering team will answer questions, provide implementation guidelines, and share benchmark data. 👉Join the community: https://lnkd.in/gZJjZWeA 🧑💻 Get started with DESILO FHE: https://lnkd.in/gZJjZWeA #DESILO #EncryptedML #HomomorphicEncryption #FHE

Our DESILO FHE Library team is in Melbourne, Australia 🇦🇺 for ASIACRYPT 2025 (Dec 8–12)! We’re excited to discover the latest results in cryptography, dive into new ideas around Fully Homomorphic Encryption (FHE), and have inspiring discussions with the community. If you’re attending, don't hesitate to say hi to Ilaria Chillotti , Damien Ligier , and Yeongmin Lee 👋 - we'd love to connect! You can find more updates, including event highlights and library news on the DESILO Discord: 📢 https://lnkd.in/gZJjZWeA #ASIACRYPT #Cryptography #FHE #HomomorphicEncryption #PrivacyTech #ResearchCommunity #TechConference #SecurityInnovation #DESILO

Great to see a live demo built with the DESILO FHE Library featured at the Collège de France seminar series. Huge thanks to Ilaria Chillotti for the terrific presentation!

We are excited to announce that the 9th HomomorphicEncryption.org Standards Meeting will be held in Seoul, South Korea on March 5-6, 2026. This meeting will take place just before the FHE.org 2026 (March 8) and RWC 2026 (March 9-11) conferences, which will both be held in Taipei, Taiwan. You can find the latest information about the upcoming #FHE Standards Meeting at https://lnkd.in/e-HWUwES Start planning your trip now! We are grateful to the local organizers from LG Electronics and DESILO Inc., including Min-Wook Jeong, Ilaria Chillotti, Damien Ligier, Myoungbee Sung, and Jun Hong Park. Looking forward to seeing many of you in Seoul! Rachel Player, Shai Halevi, Ro Cammarota, Kurt R Rohloff, Jung Hee Cheon, Kristin Lauter #homomorphic #privacy

DESILO × Cornami Announce Breakthrough in Encrypted AI Computation Our joint research with Cornami, Inc. on encrypted AI has now been officially released as a public news announcement. Led by Craig Gentry and Yongwoo Lee, this work marks meaningful progress toward practical, scalable privacy-preserving AI.   🔍 Want to dive into the details? https://lnkd.in/gV-ZGC4H #Cornami #DESILO #PrivacyAI #EncryptedAI #디사일로

Our FHE library is paving the way for real-world applications of Fully Homomorphic Encryption, from research to machine learning implementation. At Cryptography at Work for Blockchain🇫🇷 and CNCC 2025 FHE Forum🇨🇳, we showcased how the DESILO FHE library simplifies the development of privacy-preserving ML models, a key milestone in making advanced cryptography practical and accessible. This milestone reflects an important step in our roadmap toward enabling privacy-preserving AI across industries. Special thanks to our speaker Ilaria Chillotti and all event organizers for the opportunity to share our work and exchange ideas with the global cryptography community.  🔗 Explore the DESILO FHE library: https://lnkd.in/gDF9X7mB #FHE #EncryptedAI #HomomorphicEncryption #DESILO #디사일로

How can encrypted AI run efficiently without compromising privacy? Our team’s latest research takes an important step toward answering that question. We’re excited to announce that our paper on Gentry-Lee Scheme, a new computation engine that significantly accelerates homomorphic encryption operations, is now available on IACR ePrint, by Gentry and Lee. The matrix multiplication is one of the most important and time-consuming tasks in most AI tasks. Gentry-Lee Scheme introduces an optimized architecture for various-scale matrix operations, bridging the long-standing gap between theoretical FHE schemes and real-world deployment. It accelerates encrypted matrix multiplication about x80 times, and its efficiency shines for larger matrices. It achieves major efficiency improvements in encrypted computation and lays the foundation for DESILO’s encrypted AI stack that powers secure analytics and privacy-preserving intelligence. This research was authored by Craig Gentry and Yongwoo Lee, developed as part of the DESILO Inc.–Cornami, Inc. collaboration to bring scalable FHE into practice. See you at more conferences and stages ahead! 📄 Read the full paper here: https://lnkd.in/gZdPMQFr #FHE #PrivacyEnhancingTechnologies #EncryptedAI #HomomorphicEncryption #DESILO