Tumblr’s owner is striking deals with OpenAI and Midjourney for training data

Automattic, the owner of Tumblr and WordPress.com, is reportedly in talks with Midjourney and OpenAI to provide training data scraped from users' posts, according to a 404 Media report that says the deals are "imminent." The report cites internal posts describing an "initial data dump" of "all Tumblr's public post content between 2014 and 2023," which apparently included content not publicly visible on blogs. Automattic's statement published after the report says it blocks major AI platform crawlers by default and will share only public content from sites that haven't opted out. It also says it is "working directly with select AI companies" whose plans align with attribution, opt-outs, and control. It is unclear what was done with the data or whether any has been sent to either company.