Confessions to a Data Lake

[gus_massa]

> You could say that LLMs are the first technology where the medium actively invites confession.

What about ELIZA? https://en.wikipedia.org/wiki/ELIZA

[RadiozRadioz]

How does this work technically?

I am unable to go to their website because:

> "This application requires passkey with PRF extension support for secure encryption key storage. Your browser or device doesn't support these advanced features"

Is this really necessary for a product's webpage? I would understand for the application itself.

[3s]

It uses confidential computing primitives like Intel TDX and NVIDIA CC, available on the latest generations of GPUs. Secure hardware like this is a building block to enable verifiably private computation without having to trust the operator. While Confer hasn’t released the technical details yet, you can see in the web inspector that they use TDX in the backend by examining the attestation logs. This is a similar architecture to what we’ve been developing at Tinfoil (https://tinfoil.sh) if you’re curious to learn more!

[TimC123456]

There's a more-recent post on the same blog that gets into the details of how they're using the WebAuthn PRF extension to store key material, but for platforms and browsers that don't yet support the extension, you'll need a password manager that does. There's a table midway down the post with details: https://confer.to/blog/2025/12/passkey-encryption/

[viccis]

>You could say that LLMs are the first technology where the medium actively invites confession.

I don't think LLMs really constitute a medium by McLuhan's definition. A medium is an extension of man, and LLMs don't extend so much as they replace. An LLM is closer to a secretary or similar force multiplier than an extension of oneself.

[Waterluvian]

I think it absolutely is a medium based on his big picture idea of how the way we consume information influences us. A TV series or a web search or a university lecture or an evening in the stacks all deeply influence how we might study a subject and in what ways the facts take hold within us. An LLM query or “conversation” is a distinct sibling to those. And a secretary is a medium.

[recursivetree]

Neither the article nor the linked application explain how confer maintains privacy. As much as I'd like to see something like this, I'm still a bit skeptical as long as they can't explain how the technology at the heart of their product works.

[exo762]

Post is lacking in technical details. What it seems to be doing echoes the way ChatGPT is integrated into iOS - your requests are anonymized so your profile can't be (easily) built.

How can I make confer.to work on my Linux machine? Modern CPU.

[sosodev]

I’ve spent some time thinking about privacy and LLMs. I developed the impression that encryption isn’t meaningful in this space. It seems like end to end encryption only truly works when both ends are outside of the system and can manage their keys independently of it. In this case one end is the system. My message has to be decrypted for processing by the LLM. So is “end to end encryption” in this case any different than HTTPS? It doesn’t seem like it

[lisbbb]

I always said our data lake was frozen over. You could get tons of data into it, but barely anything back out.

[jmulho]

That is literally the definition of a data lake. The one you can get anything back out of is called a database.