Polyhedra and Berkeley RDI launch first production-ready zkML system

Four years after introducing the concept of zero-knowledge machine learning, Berkeley RDI and Polyhedra have unveiled a production-ready system that could redefine how trust and transparency are built into AI.

Announced today and shared with crypto.news, the new zkML technology allows developers to prove the accuracy of AI outputs without exposing sensitive underlying data or models, according to the company’s press release.

zkML

At its core, ZkML applies zero-knowledge proofs to machine learning. ZKPs are an encryption technique that allows a party to prove that a statement is true without revealing the data behind the statement.

This approach addresses trust concerns in AI, which often involves “black box” systems that lack transparency. With zkML, users can verify that their AI systems are operating as intended while maintaining privacy and compliance.

From Research to Truth

The ZkML concept was first introduced in 2020 by Polyhedra’s Chief Scientist Jiaheng Zhang and Berkeley researchers Yupeng Zhang and Dawn Song. According to the statement, at that time zkML was purely theoretical due to the high computational demands of ZKP systems.

Today, advances in zero-knowledge technology, such as Polyhedra’s Expander proof system, have made it practical to deploy zkML in real-world scenarios.

How will zkML be used?

Beyond validating AI outputs, zkML has the potential to transform how AI systems manage privacy and accountability. It enables authenticated data tagging to verify that tagged data remains accurate and unaltered, while ensuring accuracy and traceability of AI training data by facilitating data source verification. Additionally, zkML allows verification of the training process by proving that AI models are trained according to strict protocols.

Polyhedra predicts that zkML will play a key role in combining artificial intelligence with blockchain technology. This can support decentralized AI ecosystems, secure model deployment, and privacy-focused applications.

As zkML evolves, its supporters see it as a tool for building trust in AI applications without compromising privacy or security.

According to the statement, Polyhedra and Berkeley RDI plan to further expand zkML’s capabilities and make the technology accessible to developers with minimal expertise in cryptography.