Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Google is making private AI practical with homomorphic encryption

blog.google

Post reply on HN
Print view View original

241–250 of 305 posts

Re: Google is making private AI practical with homomorphic encryption

#241

Post by furkanturan »

jacquesm wrote: post #147

Private AI is perfectly practical already and it really doesn't need Google at all. Besides the word 'practical' here is really stretching things.

FHE doesn’t need Google too. Google research is investing in to it. Like we are at Belfort, various teams work on making FHE practical for organizations that want to run it for their own use cases, including on infrastructure they control themselves. Even in these setups, FHE can enable applications that would otherwise be blocked by regulatory or privacy requirements. Check out; https://belfortlabs.com/blog/belfort-partners-with-lg-on-enc... https://belfortlabs.com/blog/encrypted-fraud-detection-with-...

Re: Google is making private AI practical with homomorphic encryption

#242

Post by furkanturan »

snovv_crash wrote: post #229

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.

There are many challenges we need to solve for privacy preserving compute. Legislations, legal matters, key management, HSM like devices that has limited compute capabilities. At Belfort, we are accelerating FHE, besides exploring such uses cases for FHE. Like this one with Google, we also have publishes use cases; https://belfortlabs.com/blog/encrypted-fraud-detection-with-... https://belfortlabs.com/blog/belfort-partners-with-lg-on-enc...

Re: Google is making private AI practical with homomorphic encryption

#243

Post by Dylan16807 »

odo1242 wrote: post #134
dhx wrote: post #63

Earlier quoted context omitted.

To throw out some real and up-to-date numbers from [1] for FHE at "128-bit security level", to sort 8x 8-bit unsigned integers on the most ordinary of desktop PCs, wait 3 seconds for the result. Want to sort 32x 8-bit unsigned integers instead? Come back 34 seconds later for the result. update: also see [2] for some primitive unsigned 64-bit integer operation benchmarks with the TFHE-rs library (winner in the sorting…

It’s slightly better for LLMs because FHE is really bad at branches (it ends up essentially having to try both branches), making sorts nearly the worst possible thing to try since it’s all branches. In the case of AI most things are just addition and multiplication which can make some things faster since there aren’t as many branches. But we’re still nowhere near viability.

Sorting doesn't need any branches. For FHE you need to think like a circuit designer without indexing operations. In that situation your default sort is a sorting network, made out of nlogn conditional swaps.

Re: Google is making private AI practical with homomorphic encryption

#244

Post by HlessClaudesman »

xg15 wrote: post #227
HlessClaudesman wrote: post #223

"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.

Not even just by the authorities. This sounds like the wet dream of Google itself or any other data mining companies. You can just ship the most sensitive data around and extract signals to your hearts content - and even if it would be protected by law, you're good because hey, it's always encrypted, and the signals, ad assignments and profile markers are encrypted too! This is what I don't like about all this "I'm n…

And if they can detect "words", surely a determined attacker can reconstruct the whole thing.

Re: Google is making private AI practical with homomorphic encryption

#245

Post by furkanturan »

FloatArtifact wrote: post #2

Encryption or not, if it's on somebody else's server, it isn't yours. I don't believe Google has my best interest.

FHE is not always for using somebody else's server; but for sure you do not share keys with that somebody else.

Besides, FHE is not always about using somebody else's server. At Belfort, in addition to FHE acceleration, we also explore such uses cases; https://belfortlabs.com/blog/encrypted-fraud-detection-with-... https://belfortlabs.com/blog/belfort-partners-with-lg-on-enc...

Re: Google is making private AI practical with homomorphic encryption

#246

Post by justincormack »

Dylan16807 wrote: post #243
odo1242 wrote: post #134

Earlier quoted context omitted.

It’s slightly better for LLMs because FHE is really bad at branches (it ends up essentially having to try both branches), making sorts nearly the worst possible thing to try since it’s all branches. In the case of AI most things are just addition and multiplication which can make some things faster since there aren’t as many branches. But we’re still nowhere near viability.

Sorting doesn't need any branches. For FHE you need to think like a circuit designer without indexing operations. In that situation your default sort is a sorting network, made out of nlogn conditional swaps.

can you build a conditional swap in FHE in a small size though? FHE is always circuit like, ie no dynamic control flow anyway I think?

Re: Google is making private AI practical with homomorphic encryption

#247

Post by furkanturan »

senand wrote: post #219

I‘m new to this, but mildly skeptical. In complex agentic engineering, there are lots of intermediate steps that depend on each other. Even if you disregard the significant encryption cost, this makes it impractical. So seems like this will stay a niche.

Thanks a lot. We have some road ahead for using encrypted AI for your agentic work. Today, I would not suggest thinking that as the primary target. Instead, there are many untapped applications (e.g. inter-bank transfers, industry, healthcare) where regulations, privacy laws and compliance requirements restrict institutions from touching data. These will initially be our key enablers, and over time we hope to extend the range of applications. At Belfort we are exploring such use cases that will benefit our FHE acceleration.

Re: Google is making private AI practical with homomorphic encryption

#248

Post by furkanturan »

vkaku wrote: post #224

I really want nobody to waste cycles on my AI than me. This is useless because now I need to pay for wasted inference AND encryption cycles on top of wire encryption.

While encrypted AI chatbots seem within reach, we don’t see them as the primary target today. There is a set of untapped applications, such as inter-bank transfers, industrial use cases, and healthcare; where regulations, privacy laws, and compliance requirements prevent institutions from directly accessing or processing sensitive data.

This is where we believe Belfort can make a real difference: enabling organizations to compute on sensitive data while preserving strong privacy guarantees. And FHE doesn’t necessarily mean relying on Google, AWS, or Azure for encrypted compute. Companies can run FHE on their own infrastructure, allowing them to process their clients’ data privately while maintaining control over their systems and keys.

These applications are our initial focus, and over time, we aim to expand the range of use cases that FHE can unlock.

Re: Google is making private AI practical with homomorphic encryption

#249

Post by Dylan16807 »

justincormack wrote: post #246
Dylan16807 wrote: post #243

Earlier quoted context omitted.

Sorting doesn't need any branches. For FHE you need to think like a circuit designer without indexing operations. In that situation your default sort is a sorting network, made out of nlogn conditional swaps.

can you build a conditional swap in FHE in a small size though? FHE is always circuit like, ie no dynamic control flow anyway I think?

Once you calculate which value is smaller, the conditional swap itself is trivial. Let's say inputs A and B, comparison result is a boolean C. Outputs X and Y.

Bitwise: X = A&C | B&¬C, Y = A&¬C | B&C

Arithmetic: X = A*C + B*(1-C), Y = A*(1-C) + B*C

Edit: Or to put it another way, one of the most basic things you can make in a circuit is a multiplexer, and a conditional swap is two minimum size multiplexers next to each other.

Re: Google is making private AI practical with homomorphic encryption

#250

Post by furkanturan »

Danox wrote: post #156

The only question is do you trust Google? No…

That is the key advantage of FHE: you encrypt your data locally, using keys that you own, and never need to share the plaintext—or the keys—with anyone.

But there is more to it. Teams like us at Belfort, we are working to make FHE practical for organizations that want to deploy it for their own use cases, including on infrastructure they control themselves. Even in these setups, FHE can enable applications that would otherwise be blocked by regulatory, privacy, or compliance requirements.

Post reply on HN