Ever more websites are using #Cloudflare to block #AI scrapers. Cloudflare is still a man-in-the-middle #MITM attack on the web, but I do think people should have the ability to block the AI crap. So I now have some sympathies for using Cloudflare. What if we had real gov #eID that could be used for captchas? This requires privacy-respecting services that only see the data they need, e.g. "are you a human with an eID? yes/no". There are concerns with eIDs but in implementations not the core idea
@eighthave but a client side scrupt could automate the interaction witha real id card then automate scrappung using this id.
And no doubt people will rent the use of their id by scrapping companies...
@tuxicoman sounds like a well known problem with known solutions, for example, APIs with rate limiting, tokens, etc.
It looks you want to do the same as copyright holders. Control usage of published content. DRM in a sense.
There could be watermarks in your content you could retrieve in the AI outputs but it's fragile.
There could be legal obligations by AI companies to justify the source of their materials (we have it for food, intellectual property) and show proof of rights of use.
How can AI comapnies train of anna's archive and it's fine for them....only.
@tuxicoman yeah, I think putting legal obligations on AI companies is really something that should happen. It is funny, after all these years of trying to reduce the usage of #copyright, I think it is a good idea that AI companies have to respect copyright. The law clearly has a key role to play here.
Also I suspect those companies can embed their customers in the process.
Gmail users give other users content to Gmail.
Facebook user give picture content containing other humans to Facebook
Same for contact books in Whatsapp....
This is a real problem.
@tuxicoman that's not how copyright law works. Currently everything is copyrighted once its created, that includes emails. Just because someone forwards someone else email does not mean that they can grant a license for a text that someone else created.
@tuxicoman No, I'm not proposing DMA, just following copyright law and not caving into the AI companies or even changing the laws for them.
@tuxicoman smartcards are slow anyway, so rate limiting is built in, that would be the rock solid version. The app version could just need to implement something like that but people could hack the app.
I'm sure the renting of eID would also be regulated or even just banned, not just a free for all. For example, you're not allowed to "loan" your physical ID to your cousin who looks like you. That would be fraud.
@eighthave you mean throttling usage per "detected" same request origin (same id here)?