Only sites that have not opted out are included, and Automattic states that it asks the AI companies to remove content after a site opts out.
This page sets those terms out for readers of AI training data licensing, credits each to the party that stated it, and lists what is still unknown as of 2026-11-17.
Fact-check note: every term on this page is attributed to the party that stated it and dated; the approach is described in how we source, date and check every figure.
Automattic's AI content sharing at a glance
Automattic's own post of 2024-02-27 is the only primary source on record for this arrangement, and it states the scope and the opt-out but not the counterparties or a price.
The table records the one arrangement as Automattic states it (as of 2026-11-17).
| Date | Seller | Recipients | What is shared, as stated | Use stated | Value | Stated by | Source |
|---|---|---|---|---|---|---|---|
| 2024-02-27 | Automattic Inc. (WordPress.com, Tumblr) | select AI companies, not named by Automattic | public posts from WordPress.com and Tumblr hosted sites that have not opted out; AI companies updated about new opt-outs and asked to remove content from past sources | not stated | not disclosed | Automattic, "Protecting User Choice" | automattic.com |
Seller: Automattic Inc. (SEC CIK 0001608418, Form D filer; EDGAR, checked 2026-11-09). The parent tracker carries the same row with the recipients "not named by Automattic" and the value "n/d".
The record holds one arrangement, one program and one seller. Automattic describes the program as content sharing with select AI companies, and the table shows only fields that Automattic itself states, so every empty field means that no party has filled it.
What Automattic announced on February 27, 2024
In a post titled "Protecting User Choice", Automattic states that it shares public content hosted on WordPress.com and Tumblr with select AI companies, and only from sites that have not opted out.
Automattic's verbs are "share" and "select AI companies". This page uses the same words, because no party on record states that any payment was made, so the arrangement is not described here as a sale.
The deal record held for this page states no use of the shared content, and this page adds no use beyond Automattic's own wording. The parent tracker codes the use for this row as not stated.
The announcement is dated 2024-02-27 and sits on Automattic's own site. The record holds no separate statement by an AI company on the terms, so this page reports only what Automattic says and does not infer terms from the identity of the recipients, which is unknown.
Which WordPress.com and Tumblr content is in scope, and which is not
Automattic limits the sharing to public posts on sites it hosts on WordPress.com and Tumblr, and excludes sites whose owners have opted out.
The table lists the scope terms as Automattic states them (program record checked 2026-11-09; post dated 2024-02-27).
| Term | As Automattic states it | Source |
|---|---|---|
| Content | public WordPress.com and Tumblr posts | automattic.com |
| Eligibility | WordPress.com and Tumblr hosted sites only | automattic.com |
| Excluded | sites that have opted out | automattic.com |
| Process | opt-out setting for users; AI companies notified of new opt-outs | automattic.com |
| After an opt-out | removal requested for content from past sources | automattic.com |
| Payout terms | not published | automattic.com |
| Recipients | not named | automattic.com |
"Hosted" means the site runs on Automattic's own platforms. A site on another host falls outside the eligibility Automattic states. Automattic's record gives no statement on private content, and this page states none.
For a site owner the practical reading is narrow: the stated scope is public posts on hosted sites, with an exclusion for opted-out sites, and nothing in the record widens or narrows it further.
The site setting that keeps WordPress.com and Tumblr content out of sharing
Automattic's process for site owners is an opt-out setting: a site that uses it is left out of the content shared with AI companies, Automattic states (program record, checked 2026-11-09).
The setting is per site. Because Automattic includes only sites that have not opted out, a public hosted site on either platform is eligible until its owner acts. The program record names the process as "opt-out setting for users" and gives no deadline.
The setting acts on Automattic's own sharing. It does not reach content that other parties have already collected from the open web. How training settings on other platforms compare, and where they stop, is covered under platform settings for opting out of AI training.
Removal requests to AI companies after a site leaves the sharing
Automattic states that it updates the AI companies about new opt-outs and asks them to remove that content from past sources.
This is a request that Automattic says it makes. No AI company has stated on the record that it complies, and no source states whether removal reaches models that were already trained, so the page does not describe removal as completed.
The request is also tied to new opt-outs: the record describes Automattic updating the AI companies as sites opt out, not a one-time sweep of past deliveries.
Which removal routes reach which copies of content, and which do not, is covered under removing content from AI training datasets.
Why no AI company and no value appear in Automattic's record
Automattic names no AI company and publishes no value or payout terms, and no other source on record fills either gap as of 2026-11-17.
Other platform arrangements of 2024 differ. Reddit's S-1 of 2024-02-22 states USD 203.0 million aggregate contract value over terms of two to three years, with the counterparties unnamed.
Stack Overflow's releases name Google Cloud (2024-02-29) and OpenAI (2024-05-06) and state no value. Automattic names no counterparty and states no value, so the record cannot be used to benchmark a price or to identify who received the content.
How user terms, the FTC staff position of 2024-02-13 and company filings bear on licensing user content is compared across platforms under platforms and online communities licensing user content.
What the public record does not show about Automattic's sharing
As of 2026-11-17, no source on record states which AI companies receive the content, what they pay, whether they train on it, how much content they receive, from when, for how long, or on what exclusivity.
- Recipients: none named by Automattic or any other party on record.
- Value and payment: "not disclosed" for the value, "not published" for payout terms.
- Training use: no recipient statement on record.
- Volume: no figure on record.
- Start date and term: no statement on record.
- Exclusivity: no term on record, so the sharing is neither described as exclusive nor non-exclusive.
- Later developments: no party statement on record of a further agreement, and none on how users responded.
An Automattic statement, an AI company's statement or a filing would close any of these gaps. The page and the tracker row are updated when one appears.
What Automattic's record shows a holder weighing a license with an opt-out
The record shows how one host combined sharing with a per-site opt-out, not what another holder would be offered or paid.
Benefits
- An opt-out lets willing users stay in the sharing while others leave.
- Removal requests give a route after a change of mind.
Risks
- Changing terms to allow AI use draws FTC staff attention: the staff blog post of 2024-02-13 says retroactive, surreptitious changes to privacy terms may be unfair or deceptive.
- The record leaves open how users reacted and whether private content was touched, so a holder cannot read those effects from it.
Downsides
- No published value or payout exists to benchmark against.
- No recipient has confirmed it honors removal requests.
Alternatives
- An opt-in instead of an opt-out, so that nothing is shared without a user's action.
- Blocking crawlers only, with no sharing arrangement.
- Not licensing at all.
None of these points predicts the terms another platform or company would be offered. The weighing of these routes for a holder is set out under whether to sell or license data to AI companies.
Where to read the deal tracker, platform and opt-out pages around Automattic's sharing
This page covers Automattic's sharing only; the pages below cover every recorded deal, other platforms that license user content, crawlers and the FTC's position.
- AI data licensing deals with parties, dates and sources (the tracker)
- Forum and community archives as AI training data
- Stack Overflow AI licensing deals
- Reddit's AI data licensing deals with Google and OpenAI
- Wikimedia Enterprise AI deals
- AI crawlers and how to allow, block or charge them
- FTC rules on AI training data
- Consent and compensation for creators
A related page on the Yelp and OpenAI data licensing agreement follows in a later batch.
Frequently asked questions
Does Tumblr share user posts with AI companies?
Yes, Automattic states that it shares public Tumblr and WordPress.com content with select AI companies, from sites that have not opted out, announced 2024-02-27. The post uses the word "share", and no payment is stated by any party.
Which AI companies receive WordPress.com and Tumblr content?
Automattic names none, and no source on record does. The post says only "select AI companies", and the tracker row reads "not named by Automattic". A reader who wants a named counterparty to appear when one is disclosed can follow the deal tracker updates by email.
How much do AI companies pay Automattic?
No party has disclosed a value or payout terms. The record reads "not disclosed" for the value and "not published" for payout terms, and no party states that any payment was made.
Can a Tumblr or WordPress.com user keep posts out?
Yes, Automattic describes a site-level opt-out setting that leaves a site out of the content shared with AI companies. Automattic also states that it asks the AI companies to remove content from past sources after a new opt-out.