AI adoption has grown rapidly since ChatGpt’s breakout two years ago, but trust has not kept pace. While companies are keen to explore AI, concerns about data privacy and security have slowed progress, especially in the industry that handles sensitive information. According to Cisco’s 2025 AI Briefing: CEO Edition, 70% of CEOs are worried about network security, while 34% consider it one of the biggest hurdles for AI deployments.
Lattica, a Tel Aviv-based startup, is tackling the issue head-on with full homosexual encryption (FHE). This means that it is a groundbreaking encryption technique that allows calculations to be performed directly on encrypted data, so there is no need to publish them.
Startup Lattica raises $3.25 million to solve AI security with full homomorphic encryption
Fresh from stealth, Lattica brought encrypted AI to the cloud and raised $3.25 million in pre-seed funding to prove that FHE is no longer a theoretical concept. Ready for actual use.
The round is led by Konstantin Lomashuk’s Cyber Fund and has been supported by several prominent investors, including Polygon Network co-founders Sandeep Nailwal and the Open AGI Foundation.
FHE is nothing new. It has been called the “Holy Grail” of encryption for years. This technology allows encrypted data to be processed without decrypting. catch? It’s not practical for AI, painfully slow and resource-hunger. Lattica believes that by combining FHE with the latest AI acceleration technology, it cracked the problem.
At the heart of that solution is a platform that allows users to safely execute AI queries through hardware and the cloud they exist. This means that data remains encrypted throughout the process, including training and inference.
Founder and CEO Dr. Rotem Tsabary brings academic firepower into the effort. She holds a PhD in Lattice-Based Cryptography from the Weizmann Institute and sees similarities between machine learning operations and FHE. That overlap put her team in the upper hand to make it commercially viable against AI.
One of Lattica’s co-innovations is what it calls healing. This is a short for the homogenous encryption abstraction layer. It acts as a coupling layer between FHE applications and AI algorithms and works across custom chips such as GPUs, TPUs, CPUs, and even ASICs and FPGAs. In fact, think of it as an infrastructure that will ultimately create encrypted AI quickly enough to be used in a real setup.
“We have realized that by combining advances in hardware acceleration with software-based optimization, we can not only improve fhe efficiency to commercial viability, but we can also use it to solve the critical data dilemmas that hinder AI adoption in sensitive industries.”
The company will first focus on healthcare and financial services. This is two areas where privacy laws and regulations headaches make using AI difficult. From secure data analysis in clinical trials to encrypted financial models, Lattica bets that these industries can move through these industries overcoming the challenges of trust in AI.
As part of the launch, the team released a demos on the site and encountered the entire FHE community along with the results of their investigations. Big Point: 71% of respondents believe that the key to adoption is combining hardware and software breakthroughs. It lined up neatly with what Latica is building.
“Lattica pushes the boundaries of fully homomorphic encryption and solves one of the most important challenges in AI security,” says Konstantin Lomashuk, managing partner at Cyber Fund. “The CyberFund is proud to lead the pre-seeded round of Latica, a kind of deep technology innovation that defines the future, and we look forward to seeing Latica lead.”
Sandeep Nailwal states, “Lattica’s product-first approach fundamentally transforms the sensitive data processing of the AI ecosystem. Lattica has become a practical and scalable reality as Tsabary and her research team prove that FHE’s performance can dramatically improve and have an immediate impact on the market.
With its cloud-based services and a encryption-first approach, Lattica wants to query AI models as securely as sending secure messages. No decoding is required. No trust is necessary.
🚀Want to share the story?
Submit your stories to TechStartUps.com in front of thousands of founders, investors, PE companies, tech executives, decision makers and tech leaders.
Please attract attention
Source link
