By Richard Harris • May 9, 2025
Published on App Developer Magazine
By combining fast homomorphic encryption with trusted execution, DataKrypto launches new AI models through FHEnom for AI, enabling secure, zero-knowledge processing that protects sensitive data and model integrity without sacrificing performance.
DataKrypto launched a new solution that protects AI models and the data of businesses using them. Based on the company’s patented FHE technology, the solution, FHEnom for AI, addresses a critical security gap and delivers unprecedented AI protection.
FHEnom for AI is a zero-knowledge framework that safeguards both customized open-source AI models (adapted for specific enterprise needs) and proprietary models, while ensuring the privacy and protection of sensitive data. It defends against malicious activities, misuse, and adversarial manipulation, securing intellectual property and maintaining the confidentiality of critical datasets.
This capability establishes trust and reliability in AI-generated outputs, giving enterprises confidence in AI to make accurate, safe, and fair decisions. The solution removes barriers that prevent companies from fully embracing AI, given data privacy and model security concerns.
“While AI promises to drive significant advancements across industries, many companies have strictly limited its use for fear of their data falling into the wrong hands. Their concerns are warranted. Until now, AI has not been secure. Sensitive data, intellectual property, and the AI models companies build and train using their data have been vulnerable. DataKrypto is changing that with our new FHEnom for AI solution. By enabling fine-tuning of AI models on encrypted data, at near-plaintext speeds, we are making it safe for companies to leverage AI to innovate new solutions and solve the most complex challenges,” said Ravi Srivatsav, CEO of DataKrypto.
“DataKrypto’s new FHEnom for AI is groundbreaking in its ability to protect AI models and data while queries are processed on encrypted data, with the speed AI demands. The impact of this innovation will be far-reaching as businesses across industries can leverage AI with peace of mind that their models and data will not be subject to malicious use,” said Janakiram MSV, Principal Analyst, Janakiram & Associates.
DataKrypto launches protection for AI models: AI’s inherent insecurity
The widespread adoption of open-source and proprietary AI models has introduced critical vulnerabilities. Companies share sensitive corporate data with these models to improve decision-making, automate workflows, and unlock operational gains, but current AI tools lack robust security capabilities to protect against malicious access, data breaches, and adversarial attacks like model poisoning. These vulnerabilities allow hackers to manipulate AI outputs, compromise decision-making, and expose organizations to regulatory penalties for violating data privacy mandates such as GDPR and HIPAA.
Charting a new path with zero-knowledge AI
FHEnom for AI is a zero-knowledge AI framework that combines DataKrypto’s revolutionary FHE with Trusted Execution Environments (TEEs), establishing an end-to-end security architecture that safeguards AI model integrity and data confidentiality at every stage of training, inference, and deployment.
FHE enables direct computation on encrypted embeddings, keeping both model weights and user data protected in ciphertext throughout processing, while TEEs provide hardware-enforced isolation for secure tokenization and output within a cryptographically verified enclave. As a result, queries and AI models are blind to each other throughout the query, analysis, and output process.
By merging FHEnom’s cryptographic privacy with TEE’s tamper-resistant execution, FHEnom for AI enables a zero-knowledge AI infrastructure where sensitive data never exists in exposed form, even to the AI provider’s systems. In the event that an AI model is stolen, it is rendered useless, as it can only operate within the TEE. Because the encryption key exists only within the TEE, only users with the encryption key can see the results generated by each query.
“We address AI’s three core vulnerabilities, model security, data confidentiality, and integrity assurance, with FHE that’s fast enough to keep pace with real-time AI workflows. Speed is critical for AI’s massive data volume scale and extreme latency sensitivity. In an era where nanoseconds matter, FHEnom for AI ensures encrypted protection for models and data, delivering the scale and speed needed to fuel innovation,” said Luigi Caramico, Founder and CTO of DataKrypto.
Richard Harris
As the Publisher and Editor for App Developer Magazine, Richard has several industry recognitions and endorsements from tech companies such as Microsoft, Apple and Google for accomplishments in the mobile market. He was part of the early Google AFMA program, and also involved in the foundation of Google TV. He has been developing for mobile since 2003 and serves as CEO of Moonbeam Development, a mobile app company with 200 published titles in various markets throughout the world including the top GPS location app Chirp GPS. Richard is also the founder of LunarAds, a mobile cross-promotion and self-serv mediation network for developers. He has been a featured presenter at trade-shows and conferences, and stays active with new projects relating to mobile development.