Google's New Open-Source Compiler Makes Private AI Practical
Google Is Making Private AI Practical with Homomorphic Encryption
Google has released HEIR, an open-source compiler that lets developers run AI models on encrypted data without decrypting it, enabling private inference for tasks like recommendations and fraud detection. The tool lowers the barrier to homomorphic encryption, which previously required expert cryptographers. Google partnered with hardware accelerators and universities, and demos show single-threaded CPU latency for four applications, including a recommendation model and a hotword detector.
Servers can process ciphertexts and return encrypted results without exposing any underlying information.
- sabretooth1405
My master's thesis is on a topic in this field (Privacy Preserving ML) and from my understanding HE and other techniques have very high overheads(~10^3) on inference tasks and thus aren't very commercially viable.
- maxo133
Funny to read this, but google is currently number one anti-privacy big tech company. No other company makes it as difficult for its users using anonymization/privacy software to access their services
- meindnoch
Great, private AI, at the cost of >1000x the resource usage. Because apparently AI companies weren't already using quite enough energy to cook the planet.
The most private AI is the one running on my own hardware, not in some giant data center.
- magios
https://fhetextbook.github.io/ which i believe was linked on this website previous is what i used to learn about how this stuff works.
- Cider9986
This is the same Google that doesn't have e2ee on their password manager by default. Like WTF, it's a password manager.
- lsb
Google is making private AI practical with Gemma4 something that you can run without an Internet connection.
All of the proofs of privacy rely on us getting the math right. All of the privacy from unplugging your internet cable is there by default.
- Danox
The only question is do you trust Google? No…
- bmenrigh
Has FHE really progressed so far that it's now so efficient that doing computation on an encrypted prompt is feasible? I thought even basic operations like FHE addition were still thousands of times more complex. The only mention in the article I see is:
> But while homomorphic encryption has a nontrivial cost overhead, it shifts the capability/privacy trade-off to a question of cost. And the cost of homomorphic encryption is rapidly decreasing.
Which doesn't spell out exactly hon "nontrivial" the cost overhead still is.