Tumblr and WordPress are about to close agreements to sell user data to artificial intelligence companies OpenAI (ChatGPT) and Midjourney. The parent company of the platforms, Automattic, is about to close a deal to provide data that helps train the models of AI companies, as reported by 404 Media.
It is not yet clear what data will be included, but the article suggests that Automattic may have initially overstepped. A supposed internal post by Tumblr’s product manager, Cyle Gage, suggests that Automattic was prepared to send private or partner-related data that supposedly was not going to be included in the agreement.
Among the sensitive data found were, apparently, private entries in public blogs, deleted or suspended blogs, unanswered questions (therefore, not published), private answers, entries marked as explicit, and content from premium partner blogs.
The internal post suggests that Automattic engineers are preparing a list of post IDs that should have been excluded. “Our partners will respect all opt-out choices,” reads the end of the post. “We also plan to go a step further and regularly update any partner on individuals who have recently opted out and requested their content to be removed from past and future compilations.” It is not clear if the data has already been sent to AI companies.
