WordPress users were alarmed this week after reports by 404 Media revealed documentation showing WordPress parent company Automattic was in final negotiations to share users’ published data with AI companies OpenAI and Midjourney for training their models.
VentureBeat, as a customer of WordPress (we use their platform for publishing this article and our entire site), received this email text from CEO Nick Gernert containing an excerpt of a blog he posted today on the VIP Lobby:
News reports may have discussed our parent company Automattic’s announcement to sell data from WordPress.com and Tumblr to OpenAI and Midjourney; an initial report published in 404 Media was later picked up by The Verge.
WordPress VIP customers, I would like to assure that their data has not been disclosed through any contract signed between Automattic and themselves, nor will we ever share your personal information without explicit permission from yourself.
AI crawlers and your content: Although you cannot completely prevent AI crawlers from crawling it, reputable organizations should know not to crawl it. We provide more details on managing AI crawlers in our documentation.
Reach out directly – my email is ng@automattic.com – if I can be of any assistance in any way. Thanks again for being part of our partnership.
Gernert’s blog post and email text specifically state that “WordPress VIP customers” can rest easy knowing their data has not been shared without explicit consent – though WordPress VIP is one of three tiered options offered under the brand named WordPress — the costliest being $25,000/year designed more for enterprises rather than individuals.
Gernert left an email address in his blog post so I emailed him directly with my query about his content-sharing policy for users of WordPress’s free open-source blogging code repository WordPress.org and its hosted version WordPress.com; these latter two serve individual bloggers and small publishers respectively. As stated on its VIP site, collectively WordPress powers approximately 40% of the internet so licensing any part for AI training would represent “big f***ing deal.”
