There’s a huge difference, an absolute chasm, between non-destructive scanning, which was always an option by the way, for the purposes of putting it in a public archive

Like Anna’s Archive, or Sci-Hub. Famously legal.

If only they uploaded everything to those archives, the scanning would be perfectly fiiiiine.

You could take those downvotes as an invitation to check if those other stories (environmental, economical, societal) hold up any better.

You know how these AIs output tokens? I’m going to explain the concept with words, because that makes it easier to understand. But it means that the explanation is [ETA:]NOT quite right.

An AI has a vocabulary. The words in that vocabulary are assigned to 2 groups.

Sometimes when the AI is outputting something, it could use different words equally well. It’s not quite the same as having synonyms, cause this isn’t really about words. But let’s say you have synonyms in those different groups. At those points in the text, you can pick from one group or the other to embed a hidden pattern in the text.

Limitations are obvious. To embed the watermark, you need enough opportunities to pick “synonyms”. It won’t work for very short texts, or if the word choices are very constrained.

I’m curious if the negative effects are really as minor as they say.

Natural gas. In the EU that is classified as sustainable, green energy. Or maybe that’s only if it’s sourced from Russia. Common regulatory W.

May I ask what the use case was?

Looks like we need to regulate the regulators.

I was misremembering. Google won the precedent, when they were suing over Google Books. IA had only 1 big lawsuit and were forced to settle. The non-destructive scanning is dicey.

Copyright. They mustn’t make a copy. There is legal precedent that confirms it’s okay to transform the copy you bought into digital format. The Internet Archive relies a lot on that. I think they actually litigated it in the first place. So that’s why the copyright heads are going so absolutely apeshit. If no one’s charging you rent for using some data, then it’s “unethical”.

Boys need good, male role models.

https://jfrog.com/blog/jfrog-and-openai-collaboration-on-zero-day-security-findings/

https://huggingface.co/blog/agent-intrusion-technical-timeline

I tried to come up with a mildly funny title but… gestures helplessly at title

Sorry.

I’m not sure how much sense this makes for European citizens. What the commission wants is that Google should direct more users to European competitors rather than its own services. For example regarding price comparison sites. The major German competitor is Idealo, owned by right-wing media empire Axel Springer. Some of this seems to be only about serving the interests of our European oligarchs.

Yes, but that only puts an upper limit on the price that anyone would be willing to pay.

It’s like this: Both companies estimate how much money they expect to make from using those servers for their own purposes. In Mistral’s case, that might be a negative amount. Regardless, Microsoft obviously can put the capacity to better use. So they can offer to rent it from Mistral for more than Mistral believes it can make on its own. Basically, MS passes on some of their profits to Mistral.

Continental Europe had worse copyright laws since the 19th century. The US founding fathers were enlightenment age idealists, hence the copyright clause. European copyright laws were created by authoritarian empires. It’s a major reason why we can’t do tech. And I suspect also why Europe can’t answer to the dominance of US culture.

Legal theft like the Mickey-Mouse-Protection-Law is controversial in the US. In the EU, it goes by without comment. That ends up giving casual observers an exactly reverse impression of the legal and social situation. Same for a lot of issues.

Very minor. They are simply not illegal at present. This does not put a check on lawmakers wanting to outlaw or “regulate” them.

a) The EU extended copyright before the US. It went even further than the US by reviving previously expired copyrights.

b) The relevant copyrights are expired in most of Europe. They still exist in the NL because of a special exemption granted by lawmakers.

What would you consider sufficient to confirm the story?

That reddit comment makes no sense.

The perp has been found.

https://openai.com/index/hugging-face-model-evaluation-security-incident/