DESILO's THOR Framework Selected as Reference for Encrypted AI in Global FHE Benchmarking Suite

DESILO Inc. announced that its THOR framework, the first to enable entire Large Language Model (LLM) inference directly on encrypted data, has been added to the global Fully Homomorphic Encryption (FHE) Benchmarking Suite. This inclusion establishes a universal standard for comparing and verifying the performance of encrypted AI inference, paving the way for secure deployment of powerful AI models on sensitive data in sectors like healthcare, finance, and public administration. FHE allows computation on encrypted data without decryption, offering robust privacy protection for AI environments.

Context

DESILO Inc. has developed the THOR framework, which is the first of its kind to facilitate Large Language Model inference directly on encrypted data. Fully Homomorphic Encryption (FHE) is a technology that allows computations to be performed on encrypted data without needing to decrypt it first, thus maintaining data confidentiality. The global FHE Benchmarking Suite aims to create a standardized method for assessing the performance of various encrypted AI solutions.

Why it matters

The selection of DESILO's THOR framework as a reference for encrypted AI is significant because it sets a standard for evaluating the performance of AI models that operate on sensitive data. This development enhances the potential for secure AI applications in critical sectors such as healthcare and finance, where data privacy is paramount. By enabling AI to function on encrypted data, it addresses growing concerns over data security and privacy breaches.

Implications

The inclusion of the THOR framework in the FHE Benchmarking Suite could lead to broader acceptance and integration of encrypted AI solutions across various industries. Organizations that prioritize data security may benefit from enhanced trust in AI systems, potentially leading to increased investment in these technologies. However, the reliance on such frameworks may also require businesses to adapt their existing infrastructures to accommodate new standards.

What to watch

In the near term, stakeholders in sectors that handle sensitive data will likely monitor the adoption of the THOR framework and its impact on AI deployment strategies. The effectiveness of the benchmarking suite in evaluating different encrypted AI models will also be closely observed. Additionally, advancements in FHE technology may lead to further innovations in secure AI applications.