In a development that might catch Tumblr users off guard, Automattic, the parent company of both Tumblr and WordPress.com, is reportedly discussing with AI firms Midjourney and OpenAI. A recent report by 404 Media, based on information from an anonymous source inside Automattic, indicates that a deal involving Tumblr users’ posts as potential training data for AI systems could be imminent. This news has stirred the online community, raising questions about data privacy and the ethical use of user-generated content.
Opt-out option amid data scrape concerns
Automattic is preparing to introduce a new setting this coming Wednesday, enabling users to choose whether to share their data with third parties, including AI companies. This development follows internal revelations suggesting Automattic might have already undertaken a significant data scraping initiative. According to these internal posts, the company has collected an “initial data dump” that includes all public Tumblr posts from 2014 to 2023. This collection reportedly also encompasses content that was not intended to be publicly accessible. The exact use of this data, and whether any of it has been shared with Midjourney or OpenAI, remains a matter of speculation, as none of the involved companies have made any official statements yet.
The balancing act: Innovation vs user rights
The dilemma faced by Automattic is not unique in today’s digital landscape. Numerous companies have negotiated with AI tool developers, providing them with training data. This trend includes high-profile deals like Reddit’s US$60 million annual agreement with Google and Shutterstock’s collaboration with OpenAI to use its photo library. However, such arrangements are increasingly being scrutinised, particularly by the creative communities on platforms like Tumblr. Artists and writers have opposed using their work for AI training without consent. This pushback forces companies to tread carefully, balancing the allure of AI innovation against the need to respect user rights and privacy. This challenge has been evident in the recent backlash against online platforms like DeviantArt, which have experimented with AI technologies.
Automattic’s quest for Tumblr’s profitability
At the heart of these discussions is Automattic’s ongoing struggle to find a profitable model for Tumblr, which it acquired from Verizon in 2019. Despite the success of its other ventures, such as WordPress.com and WordPress VIP, monetising Tumblr has proven difficult. Last year, Automattic announced a decision to downscale its ambitions for Tumblr. The details and potential financial benefits of the discussed deal with Midjourney and OpenAI are still unknown. However, the implications of such a partnership could be significant, not only for Automattic’s revenue streams but also for the broader discourse on data privacy and the ethical use of AI.
Tumblr’s user base and the wider online community await further developments with keen interest. The outcome of these talks could set a precedent for how user data is leveraged in the burgeoning field of AI, potentially reshaping the landscape of online content creation and data privacy.
