Enveil, the pioneering data security company protecting Data in Use, announced the release of its encrypted machine learning product, ZeroReveal Machine Learning (ZeroReveal ML), the first adaptable, market-ready solution allowing organizations to process data against an encrypted machine learning model.
Building on the success of its ZeroReveal Search solution, Enveil ZeroReveal ML fundamentally changes the paradigm of secure data usage by allowing organizations to enable advanced decisioning through collaborative and federated machine learning in a secure and private capacity. The product expansion comes on the heels of the company’s $10 Million Series A funding announcement.
ZeroReveal ML ensures that machine learning models, and the associated results, remain encrypted throughout the entire evaluation lifecycle. This enables enterprises to leverage machine learning models to securely and privately derive insights from data sources across jurisdictional, third-party, and organizational boundaries, even when using highly sensitive or proprietary models, including those trained on regulated data.
The ability to securely evaluate a machine learning model against distributed intra- and inter-organizational data sources opens the door to a number of use cases in the financial services, retail, manufacturing, healthcare, and government verticals where enhanced risk-based decision making drives positive business outcomes.
“Organizations increasingly rely on machine learning to generate insights, make predictions, and improve decisioning,” said Dr. Ellison Anne Williams, Founder and CEO of Enveil.
“While machine learning can be an effective tool within an organization’s walls, when teams are able to securely evaluate their models over multiple third-party data sources by collaborating with external organizations, they significantly broaden their capacity to derive insights for improved outcomes.”
ZeroReveal ML delivers encrypted machine learning capabilities by focusing on the source of the value: the model itself and the insights derived. Trained machine learning models are valuable — and vulnerable.
Sensitive or regulated data is often tied up in the model itself and exposure can reveal this sensitive data as well as an organization’s interests, intentions, or Intellectual Property.
Machine learning models are organizational assets and should be protected with the same vigilance as any other type of Intellectual Property or sensitive data. By keeping the model encrypted during processing, ZeroReveal ML ensures this critical information is never revealed.
“Trained machine learning models need to be protected, which is why ZeroReveal ML focuses entirely on the security of the model itself and its evaluation,” said Dr. Ryan Carr, CTO and Vice President of Engineering at Enveil.
“Attackers are waiting to reverse engineer machine learning models and extract information about an organization, including training data which may contain PII, IP, or other sensitive material. We preserve the value of the model without that risk of exposure.”
Powered by advances in homomorphic encryption, ZeroReveal ML utilizes the same foundational ZeroReveal Compute Fabric as Enveil’s award-winning ZeroReveal Search capabilities to deploy above the data, allowing customers to leverage existing storage methods and access control mechanisms.
Organizations and jurisdictions can evaluate encrypted machine learning models across multiple datasets without moving or pooling data in a single location or requiring data standardization.
As the only NIAP Common Criteria and CSfC-certified Data in Use security company, Enveil continues to deliver business-enabling functionality and nation-state level protection to the global marketplace.