Comment by JohnFen | Hacker News Reader

Comment by JohnFen | Hacker News Reader

Parent Full thread

You take a derisive tone, but there is a violation of expectations here for a lot of people.

All of the stuff I've put on the web, I put there for actual people to use. Crawlers and scrapers came around, and while I certainly didn't like it or approve of it, it was something I put up with. The only defense against it was to stop putting things on the web entirely, which seemed like an overreaction.

Now, however, the use of that data to train AI (something that I consider actively harmful to society in general and don't want to support in any way), is a degree too far in the pot that's been slowly coming to a boil.

While I do want to give back and contribute to the larger body of work available to people, I want it to be available to actual people. I don't want it to be available for training AI.

I don't see why that's such a ridiculous stance. However, it's not legally possible to both make a work available to the public and prevent that work from being available for other uses. That's a real shame, and that the only protection available to me is to no longer have the works available publicly, it's loss to everybody.

Is... any of that coded in law somewhere or illegal?

"I don't want it to be available for training AI" is a perfectly reasonable personal preference. We can discuss whether it's selfish or not, whether others agree or not, etc.

But this is a lawsuit. What makes it illegal for OpenAI to use your content? Like, is there some license you've put up on your content that disallows it? Is there anything that's relevant to the case at hand?

> https://www.ftc.gov/business-guidance/privacy-security/child...

> The Children’s Online Privacy Protection Act (COPPA) gives parents control over what information websites can collect from their kids. The COPPA Rule puts additional protections in place and streamlines other procedures that companies covered by the rule need to follow. The COPPA FAQs can help keep your company COPPA compliant. Learn about the COPPA Safe Harbor Program and about organizations the FTC has approved to implement safe harbor programs. You can also get information about ways to get verifiable parental consent– including new methods the Commission has approved – and the process for seeking approval for new methods.

There's one. There are similar laws in Europe.

The AI scraped, utilized, kept, keeps... data on children. Parents did not consent. Children /can't/ consent.

The AI use of such data is not only illegal but could end up with people being jailed over it.

I assume the parents consented to have that information on a website, or it couldn’t have been scraped in the first place. They may be suing the wrong party here.

The kids could have done so without the consent of their guardian. It cannot be used, unless consent is given beforehand.

I never said it was illegal. I was responding to the implication that because you put something publicly on the web, it's somehow ridiculous to object to certain uses of it.

As to the legality, that remains unknown until a court makes a ruling. I suspect that this lawsuit will go nowhere, but I'm just speculating along with everyone else.

Of course as the data’s owner you can object to anything. Who is to say that you are not allowed to hate certain practice?

On the other hand, you can’t prevent people from using your data if you put it out in public without condition (in the form of some acceptable licenses).

As an analogy, if you put a picture of your living room in the sidewalk, you can’t prevent me from looking at it, study it, when I walk by. I may even benefit from it by copying your style or decoration. You may however cover it, with a warning. I’d clearly violate your terms if I still look at it against your will, although I may not violate any laws.

The problem is that a lot of that public data is protected by copyright.

It’s not obvious to me that using public data to train AI is a violation of copyright. Many are claiming it is, but as far as I know the courts haven’t weighed in on it yet.

One could argue copyright, but you’d have to prove this was copyright infringement. Which I don’t believe the law has caught up with the answer to that.

Medical data and data about children have very specific protections. The article says these were accessed without permission. If a lab tech dropped my lab results on the floor and you picked them up, it would not give you the right to store and use my information.

How were they accessed without permission if they were protected? Is that the fault of the AI folks, or those sworn to protect the information? Unfortunate and bad, but if it's up there in the open I'm more worried about the fact that's it's available than the fact that an AI scraped it. If an AI can figure it out, so can multiple bad actors.

The data can have been leaked illegally and put on the internet. That doesn't mean you can use it.

As an EU citizen, some random company can't use PII related to me if I don't give consent or revoke my consent. They'll have to remove it or face a huge fine. The law doesn't care about the cost for your company to comply.

What makes it legal? If you find code on GitHub without a license, then this does not mean "public domain". It means "the author has given permission for you to view this, but nothing more". Arguably using it for training counts as "more".

So my question would be "what license entitles OpenAi to use my content"? It could be that as part of the user agreement of the website you gave some rights to the website and they sold it on to OpenAi, could be fair use, could be "does not count as 'more'",... .

Bad analogy: For money laundering we have similar rules. If you suddenly appear with a large amount of money/data, then the obligation is on you to show that it is clean, not on others to show that it is dirty. You can disagree with that (innocent until proven guilty etc.), but it is not clear cut which way around things should be.

> it's not legally possible to both make a work available to the public and prevent that work from being available for other uses

That’s bullshit. Copyright still applies to property available to the public. Maybe you are confused between “available to the public” (access) and “in the public domain” (copyright).

Listening to a song on the radio doesn’t give you the right to copy it and sell it.

You can say the same for other forms of IP. Trademarks, logos, etc are all “available” to the public, but they are still protected IP.

When you post to a public website, you are giving access to that content , and (usually) legally giving rights to the website to publish your content. Those rights don’t transfer to a 3rd party.

Reply on news.ycombinator.com