Hacker News Reader

Comment by jsnell | Hacker News Reader

Your data is not being used to train an LLM even if you click "yes" to this.

And you know this how?

To be fair, can you even know they won't do it if you click "no"?

Disclaimer, I work at Google, but I recommend reading the ToS for Google consumer and business services, and I believe the short version of the policies on this have been documented on the Google blog several times. I can't comment on the policy directly, but encourage you to research it, there's a pretty straightforward answer.

Separately, the industry standard elsewhere appears to be that if you're a free user your data is used for training, and if you're a paying user it's not.

It's so straightforward that you can't comment on it without risking retaliation?

Don't Talk to the Police. https://www.youtube.com/watch?v=d-7o9xYp7eE

I am either reading the wrong ToS, or not getting it [1]:

    Permission to use your content
    [...]
    This license is for the limited purpose of:
    [...]
    - developing new technologies and services for Google consistent with these terms

The last item sounds like Google can use my content to train an LLM, no?

[1] https://policies.google.com/terms?hl=en-US#toc-content

Do you have a free or paid account? What are the terms for paid accounts?

I had assumed that the GP comment [1] was about both paid and free accounts.

[1] https://news.ycombinator.com/item?id=44592213

Workspace has published policies about how the customer's data may be used: https://support.google.com/a/answer/15706919

> Your content is not used for any other customers. Your content is not human reviewed or used for Generative AI model training outside your domain without permission.

And I think the default assumption has to be that those policies are actually followed. If you think they aren't, the burden of proof is on you to show they aren't.

What the OP of that comment and yourself have missed is that Google is not trustworthy enough to give it permission, which the user you're responding to pretty much suggested as a solution (Google should be trustworthy to be given a one-off permission).

The amount of times Google has been caught out doing things they shouldn't be means that no, no one has to prove that it's a reasonable assumption that Google follows policies, not even their own.

It's for you to prove that they can be trusted (good luck with that!).

In that case what do you care if you click yes or no? If you don't trust Google to abide by their own terms what makes you think they'll respect your "no"? Surely at that level of trust you would just stop using their product?

And meta hasn’t pirated the world’s library… it’s business as usual.

Reply on news.ycombinator.com