Fetching from the wire…
01
02
03
04
05
06
07
08
Infra2026-08-15 · source-backed
Announced August 14, HEIR (Homomorphic Encryption Intermediate Representation) converts pretrained AI models to operate directly on homomorphically encrypted inputs, so the server never sees plaintext. Demonstrations cover a deep-learning recommender with Belfort Labs, LG and NYU, credit-card fraud detection, Kitsune network intrusion detection, and audio hotword detection. Google Google concedes nontrivial cost overhead and quotes single-threaded CPU latency, which is the honest caveat. The stated goal is a one-click path to encrypted inference for non-experts, and that's the part that would change what regulated industries can ship.
Each link below shares sources, entities, or timing with this story.
HEIR built by Google / Shared entity: Google / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (HEIR built by Google); both cover Google; overlapping topics (cost, google, model).
HEIR built by Google / Shared entity: Google / Same source domain / Shared topic / Earlier coverage
Linked by a graph relationship (HEIR built by Google); both cover Google; reported by the same outlet (blog.google).
Linked by a graph relationship (HEIR built by Google); both cover Google; reported by the same outlet (blog.google).
Linked by a graph relationship (HEIR built by Google); both cover Google; reported by the same outlet (blog.google).
Linked by a graph relationship (HEIR built by Google); both cover Google; reported by the same outlet (blog.google).
HEIR built by Google / Shared entity: Google / Same source domain / Earlier coverage / Tension
Linked by a graph relationship (HEIR built by Google); both cover Google; reported by the same outlet (blog.google).
HEIR built by Google / Shared entity: Google / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (HEIR built by Google); both cover Google; overlapping topics (cost, google).
Linked by a graph relationship (HEIR built by Google); both cover Google; overlapping topics (change, google).