Please be aware that the "opt-out" choice is just a way to try to appease people. But Tumblr has not been transparent about when has data been sold and shared with AI companies, and there are sources that confirm that data has already been shared before the toggle was even provided to users.
Also, it seems to include data they should not have been able to give under any circumstance, including that of deactivated blogs, private messages and conversations, stuff from private blogs, and so on.
Do not believe that "AI companies will honor the "opt-out request retroactively". Once they've got their hands on your data (and they have), they won't be "honoring" an opt-out option retroactively. There is no way to confirm or deny what data do they have: The fact they are completely opaque on what do they currently "own" and have, means that they can do whatever they want with it. How can you prove they have your data if they don't give everyone free access to see what they've stolen already?
So, yeah, opt out of data sharing, but be aware that this isn't stopping anyone from taking your data. They already have been taking it, before you were given that option. Go and go to Tumblr's Suppport and leave your Feedback on this (politely, but firmly- not everyone in the company is responsible for this.)
Finally: Opt out is not good under any circumstance. Deactivated people can't opt out. People who have lost their passwords can't opt out. People who can't access internet or computers can't opt out. People who had their content reposted can't opt out. Dead people can't opt out. When DeviantArt released their AI image generator, saying that it wasn't trained on people who didn't consent to it, it was proven it could easily replicate the styles of people who had passed away, as seen here. So, yeah. AI companies cannot be trusted to have any sort of respect for people's data and content, because this entire thing is just a data laundering scheme.
Please do reblog for awareness.
Time to hit them with the GDPR bat.
I'm coming for you, Tumblr.
I wonder if GDPR would allow us to request our data from these training companies. The requests would be impossible to service, which is exactly why I’d love if if they were mandated to comply.
The big issue at hand is if that data is truly personal data. If everything is anonymized, there's not much you can do from a GDPR standpoint, however, our old frenemy copyright *might* be able to help
