Privacy tech: FHE and ZK demos

Fully homomorphic encryption is a way to send locked data to another computer, let it do math on the lockbox, and get back an encrypted result. The United States National Institute of Standards and Technology says the method supports computing functions over encrypted data without the secret key. (csrc.nist.gov) Zero-knowledge proofs solve a different problem: proving a claim without handing over the underlying data. Ethereum Community Conference, known as EthCC, has framed zero-knowledge proofs as a core privacy tool for identity, healthcare, and machine-learning applications in past conference sessions. (ethcc.io 1) (ethcc.io 2) The fully homomorphic encryption idea dates to 1978, when Ronald Rivest, Adi Shamir, and Leonard Adleman posed the problem, and to 2009, when Craig Gentry published the first candidate construction. National Institute of Standards and Technology slides from September 25, 2024, use that timeline in an overview of the field. (csrc.nist.gov) The pitch is simple: data usually stays protected while stored and while moving across networks, but it often has to be exposed during processing. IBM says fully homomorphic encryption aims to keep data encrypted at rest, in transit, and during computation. (ibm.com) That is why artificial intelligence companies keep pointing to it. National Institute of Standards and Technology materials from its 2024 Workshop on Privacy-Enhancing Cryptography say current schemes can support privacy-preserving artificial intelligence, including large language model inference in some settings, though efficiency remains a central issue. (csrc.nist.gov) Microsoft and IBM both maintain public tooling and demos that show the field has moved beyond theory. Microsoft publishes the open-source Microsoft SEAL library, and IBM runs an “AI on Encrypted Data via FHE” project page describing encrypted model inference and training workflows. (microsoft.com) (ibm.com) Zero-knowledge proofs are showing up in a parallel track around digital identity. Cointelegraph reported on December 24, 2025, that selective disclosure and zero-knowledge proofs were becoming a privacy-first option for decentralized identity systems as crypto and government-backed identity efforts converged. (cointelegraph.com) In a genomic setting, that means a person could prove something about a DNA record without posting the raw sequence on a blockchain. EthCC’s archived healthcare session described zero-knowledge systems as a way to verify patient-related claims while limiting direct exposure of sensitive medical data. (ethcc.io) The two approaches are not interchangeable. Fully homomorphic encryption is for computing on hidden data, while zero-knowledge proofs are for convincing someone a statement is true without revealing the evidence. (csrc.nist.gov) (ethcc.io) The recent demos put those tools in the same conversation: one for private computation, one for private verification. That pairing is becoming a standard blueprint for systems that want to use sensitive data without exposing it in the clear. (csrc.nist.gov) (ibm.com)