It feels as though this week starts with some little victories, and those are always good things to share. I’ll start with one of the stories I featured very recently: the move by Meta to use people’s posts across its various platforms, including Instagram and Facebook, for AI training. If you recall, users of those platforms had to jump through considerable hoops—from finding the notification in the first place to composing an essay explaining why they didn’t want their creative output to train AI—to withdraw from the scheme.
Now Meta has been told to, at the very least, pause their plans in the European Union and UK. The demand is being made under the General Data Protection Regulation (GDPR). Meta had argued that they had a legal justification to use people’s content to train their AI under the GDPR processing ground of “legitimate interest.” The EU’s Data Protection Commission (DPC) and UK’s Information Commissioner’s Office (ICO) disagreed. At the heart of their objection is Meta’s lack of transparency and the lengths it made people go to in order to opt out. Although it claims that anyone who asked was granted an opt-out, it was by no means clear that this would be the case when they requested people’s essays. And the notification of the new policy was hidden in the forest of regular notifications.
Meta is, needless to say, disappointed and has wheeled out that old favorite accusation of stifling innovation. And it will, inevitably, be back with the policy in a different form. In the meanwhile, users outside of the protection of GDPR will continue to be affected and need to make a request to opt out. For more details, you can read about Meta's stance here.
Talking of wins, I came across a post in the Wide for the Win group that purports to be from a representative of Barnes and Noble Press. It addressed the removal of paperback titles from the BN Press website, claiming that an exercise to cut down on the number of redundant titles on the site had been undertaken, and apologizing that several indie titles had been caught in this inadvertently. The post claimed those titles mistakenly removed were being restored. The poster suggests contacting [email protected] with a list of affected ISBNs if your title has been impacted.
Unfortunately there is no indication of any such consideration or protection from Meta using posts for AI training for Australians! We are fair game for Meta!