Migliaccio & Rathod LLP is investigating whether Tumblr users’ posts and other content may have been collected or shared with artificial-intelligence companies before users were given meaningful notice or an opportunity to prevent that use.
In February 2024, Tumblr parent company Automattic announced that it was working with select AI companies and introduced settings allowing users to prevent their public Tumblr content from being shared with third-party partners, including companies that train AI models. Automattic stated that only public content from users who had not opted out would be shared under those partnerships.
However, reporting based on internal Tumblr documents raised questions about what information may have been compiled before those protections were put in place.
Reports Concerning Tumblr Content
Internal materials reported by 404 Media described an “initial data dump” prepared in connection with OpenAI and Midjourney that allegedly included categories of content that were not supposed to be included.
The reported categories included:
- private posts on otherwise public blogs;
- posts from deleted or suspended blogs;
- unanswered asks;
- private answers;
- mature or explicit posts; and
- certain content associated with premium partner blogs.
The reporting did not establish that all of this material was ultimately transmitted to an AI company. Migliaccio & Rathod LLP is investigating what content was actually shared and whether any private, deleted, or otherwise restricted material left Tumblr’s systems.
Tumblr’s current privacy controls state that private posts, drafts, messages, unpublished asks, posts from deleted blogs, and certain other categories are excluded from third-party sharing.
What We Are Investigating
We are investigating whether:
- Tumblr content was provided to AI companies before users were notified;
- older posts were included in AI-training datasets without separate consent;
- private, deleted, suspended, or otherwise nonpublic material was actually transferred;
- users were automatically eligible for third-party sharing unless they opted out;
- users received adequate notice before their content became available for AI-related uses;
- users who later opted out could prevent previously shared content from being used; and
- Tumblr users received any compensation or other benefit from the use of their content in AI development.
Automattic states that when users later opt out, it will notify its partners and ask them to remove the affected content from past sources and future training.
That raises an additional question about whether content already delivered to or incorporated by an AI company can actually be removed after the fact.
Potential Claims May Include:
- unfair or deceptive trade practices;
- privacy-related claims;
- breach of contract;
- breach of the implied covenant of good faith and fair dealing;
- unjust enrichment; and
- failure to adequately disclose the commercial reuse of user-created content.
Signs You May Be Affected:
- You maintained a Tumblr account before February 2024;
- you posted original artwork, photography, writing, videos, or other creative material;
- you maintained private posts or private answers;
- you deleted a Tumblr blog or substantial amounts of Tumblr content;
- you used asks, submissions, or other features containing material that was not publicly posted;
- you did not know Tumblr content could be shared with AI-training partners;
- you discovered the Prevent Third-Party Sharing setting only after it was introduced;
- you later opted out of third-party sharing; or
- you are concerned that content posted years earlier may already have been included in an AI-related dataset.
If you believe your Tumblr content may have been collected or shared for AI training without your informed consent, we would like to hear from you.
Please complete the contact form on this page, send us an email at [email protected], or give us a call.