Examining Automattic’s AI Access Policy: A Closer Look at the Scrutiny
In recent news, there have been unconfirmed reports of Google entering into a content licensing agreement with Reddit for training its AI. Following this, 404 Media has claimed that Automattic, the company behind popular platforms Tumblr and WordPress.com, is also set to sell users’ content to Midjourney and OpenAI. This potential partnership between Automattic and AI companies mirrors the extended partnership that Shutterstock entered into with OpenAI last year.
The claims made by 404 Media are based on insider information, which they claim is backed up by documentation. To support their claims, 404 Media quoted Tumblr Product Manager Cyle Gage, who reported on an internal message board about the status of the initial data collection process and how it included content that should not have been collected.
While 404 Media has provided quotes from an internal source, they have not provided specific proof such as screenshots of conversations or access to source materials to validate their claims fully. Additionally, 404 Media refers to user content as “users’ data,” which can be easily misconstrued as personally identifiable information or credit card information. However, it is important to note that the content being discussed in the article is content that is already publicly available.
In response to these claims, Automattic released a statement within a few hours of 404 Media’s article going up. The statement describes Automattic’s position on content distribution and the rights of all users on WordPress.com and Tumblr to opt out of their public content being included in data shared with AI partners.
Automattic argues that AI regulation and legislation do not yet exist, and as a result, they are proactively taking steps to provide users with additional methods of controlling how and where their content is made available. They are creating a pathway for AI partners to gain streamlined access to the content users are open to sharing while also taking steps to remove access to content that users no longer want to be shared. It is worth noting that the content in question is already available to AI companies as it is publicly crawlable, and content deals would only make it more accessible and manageable.
Automattic has published “Protecting User Choice” to emphasize their commitment to user control over their content. The document outlines several measures taken by Automattic to discourage search engine indexing and discourage crawling by AI companies. However, it is important to note that Automattic will only share public content hosted on WordPress.com and Tumblr from sites that have not opted out.
The article also hints at future deals, stating that Automattic is “working directly with select AI companies as long as their plans align with what our community cares about: attribution, opt-outs, and control.” Automattic plans to regularly update partners about users who have newly opted out and request the removal of their content from past sources and future training.
In a statement shared with the WordPress community, Josepha Haden Chomphosy, Executive Director of WordPress, confirmed that the WordPress project is not involved in selling user data or content for AI training purposes. This stance has been consistent throughout the long history of WordPress.
It is important to note that WordPress.org users are not affected by Automattic’s AI access policy. Chenda Ngak, Head of Communications at Automattic, has been reached out to for comment on the matter.
This news comes at an interesting time as Automattic has been struggling to make Tumblr profitable since acquiring it in 2019. Last year, it was revealed that Tumblr is losing $30 million each year.
In conclusion, the scrutiny surrounding Automattic’s AI access policy raises questions about the potential sale of users’ content to AI companies. While 404 Media claims to have insider information, they have not provided concrete evidence to support their claims. Automattic, on the other hand, has released a statement outlining their commitment to user control and providing additional methods of controlling content distribution. It remains to be seen how this potential partnership with AI companies will unfold and what impact it will have on users of Tumblr and WordPress.com.
