Dropsitenews published a list of websites Facebook uses to train its AI on. Multiple Lemmy instances are on the list as noticed by user BlueAEther
Hexbear is on there too. Also Facebook is very interested in people uploading their massive dongs to lemmynsfw.
Full article here.
Link to the full leaked list download: Meta leaked list pdf
These guys don’t get that the scrapers are just going to dump their piddly little text into /dev/null. And that all they are accomplishing is making other humans hate their posts while doing absolutely nothing to poison the llms.
You can’t poison a data set of this size with a few hundred stupid comments.
All they are really going to accomplish just getting blocked by people who agree with their main point.
Or, we can look for mitigations, instead of dismissing concerns. Common enemies of Freedom 🤝?
Sure, you can look for mitigations. In the course of looking for mitigations, wouldn’t it be nice if someone let you know that the idea you’d come up with as a mitigation was not going to work?
Then let’s look for another! Whta do you propose?
I’ve given my suggestion in other comments in this thread. In short: if you don’t want your comments to be seen by all, then don’t post them on a public forum that uses an open protocol specifically designed to broadcast your comments to everyone who cares to listen. Perhaps use some closed-off forum instead, preferably run by a large and litigious company that guards its possessions jealously.
Ok, so get illegally scraped and copyright violated, got it boss.
Got any citations about this being illegal? If it is then the whole ActivityPub protocol is in trouble.
[Insert any copyright law you’re domicile to] Consult your lawyer about copyright violations of federated content. I am not yours to violate.
Have there been any relevant lawsuits you could point me to? Vaguely waving in the air and declaring “copyright” is not helpful.
They’re just using very simple scrapers that don’t have any knowledge about how the site operates. The simplest counter would probably be using Anubis on the web interface.
I wouldn’t mind waiting 2-3 seconds when first loading the site and mobile apps would remain unaffected since they use the API.