Remix.run Logo
pards 2 days ago

> Note: No generative AI was used to create this content. This page is only intended for human consumption and is NOT allowed to be used for machine training including but not limited to LLMs. (why?)

I love this. I wonder if it is effective, or if there's any legal recourse should the LLMs ignore it.

blitzar 2 days ago | parent | next [-]

When your LLM starts caveating things with "No generative AI was used to create this content. This page is only intended for human consumption and is NOT allowed to be used for machine training including but not limited to LLMs." you know they ignored the request.

diggan 2 days ago | parent [-]

We're almost there. If you train models on the output of Llama, they (try to) force you to use a particular name for example. As Meta starts to squeeze Llama (developer/research) users eventually, I'm sure they'll try to legally prevent you fully from doing so unless you use their (hypothetical future) platform/service:

> If you use the Llama Materials to create, train, fine tune, or otherwise improve an AI model, which is distributed or made available, you shall also include “Llama 3” at the beginning of any such AI model name.

https://www.llama.com/llama3/license/

blitzar 2 days ago | parent [-]

I wonder if they can train in something like the name so deeply that you can't untrained it without torturing the model so badly it just spews out garbage.

diggan 2 days ago | parent | prev | next [-]

Depends on the jurisdiction obviously (seems in this case Sweden/EU), but I don't think the author could blanket ban it like that, as research organisations and cultural heritage institutions have exceptions for example. I think news organizations and archivists have exceptions too, but less sure about that.

Besides, I think for it to actually have any effect, it would have to be in a machine-readable format, I'm not sure placing a notice like that in the middle of a HTML document is good enough.

> Art. 4(3) applies only on condition that right holders have not expressly reserved their rights “in an appropriate manner, such as machine-readable means in the case of content made publicly available online”. According to Recital 18, “it should only be considered appropriate to reserve those rights by the use of machine-readable means, including metadata and terms and conditions of a website or a service. […] In other cases, it can be appropriate to reserve the rights by other means, such as contractual agreements or a unilateral declaration.” In other words, Art. 4 right holders may effectively prohibit text and data mining for commercial uses by adding robot.txt type metadata to their content online.

https://copyrightblog.kluweriplaw.com/2019/07/24/the-new-cop...

But maybe the author already have the same notice in a machine readable and of course I'm not a lawyer or anything, so this is just guessing based on what I've learned about the regulations that affect me personally in Spain/EU.

Bengalilol 2 days ago | parent | prev | next [-]

https://blog.alexewerlof.com/robots.txt

Gives a small hint. It should be respected and effective.

arealaccount a day ago | parent | prev | next [-]

I enjoyed this notice coupled with the presumably unlicensed Family Guy image below it

CaptainFever a day ago | parent | prev | next [-]

I guess I'm not reading the article then, since I'm not human.

dangus 2 days ago | parent | prev [-]

I do not love this. I stopped reading right there. I’m not giving losers who are too lazy to write their own blog posts with their own words my engagement.

The hypocrisy of it is astoundingly obvious. The author used AI trained on stolen content to help write their blog post but then denies “the next author” from benefitting in the same way. If you love AI so much you should let it steal your content and train on it.

Classic “pull the ladder up behind me” mentality.

Atreiden a day ago | parent [-]

I think you missed the leading "No" in the sentence, "No generative AI was used to create this content"

dangus 3 hours ago | parent [-]

@_@ well that explains a lot, whoops!