Tumblr Selling Some User Content to AI Companies Without Consent
-
Tumblr's parent company Automatic is selling user data to AI companies like OpenAI and Midjourney to train their models.
-
The data being provided includes some private, deleted, or mature content that was mistakenly included, despite rules against this.
-
ChatGPT will now be exposed to lots of Tumblr fanfiction content like "Fawnlock" (Sherlock/Watson as part deer).
-
Reddit also has a deal providing user data to Google AI, while Facebook/Instagram data trains Meta's internal AI.
-
Some users feel uncomfortable about their personal content being used this way without consent, unlike professional writers.