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.
sabretooth1405
So much inefficiency just to run it on someone else's untrusted hardware. Private AI is already possible today with local open-weight models running on hardware you control.
Homomorphic encryption is cool technology, but I'm really not sure what problem it solves.
snovv_crash
I think you folks are reading too much into it. I think the people working on FHE need to publish an AI-oriented pitch to retain funding from AI-pilled execs. Must be the same case with the golang post few days ago.
never_inline
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.
meindnoch
This is the same Google that doesn't have e2ee on their password manager by default. Like WTF, it's a password manager.
Cider9986
"hotword" detection, is this basically sniffing encrypted data for words the authorities don't want you to speak?
Seems counterproductive to their stated aim of privacy.
HlessClaudesman
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.
lsb
It seems to me this tech still presumes the data sits in data warehouses, which i don't like to start with. The tech succeeds in packing the data in identical black boxes, so they all seem equal to the map-reduce function that runs over them. A separate identification layer knows which of the boxes is yours. But who knows, maybe from the results you can be fingerprinted anyway. As in the game, how many questions before you can guess the thing I'm thinking of? Answer: not that many.
childintime
Making homomorphic encryption practical for AI workloads is a massive step forward for privacy-preserving architectures. Brilliant milestone.
myshapeprotocol
https://fhetextbook.github.io/ which i believe was linked on this website previous is what i used to learn about how this stuff works.
comments (10)
sabretooth1405
Homomorphic encryption is cool technology, but I'm really not sure what problem it solves.
snovv_crash
never_inline
The most private AI is the one running on my own hardware, not in some giant data center.
meindnoch
Cider9986
Seems counterproductive to their stated aim of privacy.
HlessClaudesman
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.
lsb
childintime
myshapeprotocol
magios