For about a year, Tyler Johnston has been collecting public information about the inner workings of OpenAI, and for the past month, he’s been working on a report to help the public understand and visualize it.
That report, dubbed The OpenAI Files, is out today. It’s a collaboration between the Midas Project and the Tech Oversight Project, two nonprofit tech watchdog organizations, and it’s billed as the “most comprehensive collection to date of documented concerns with governance practices, leadership integrity, and organizational culture at OpenAI.”
The interactive site, which amounts to more than 50 pages and over 10,000 words, chronicles OpenAI’s evolution from nonprofit research lab to moneymaking household name, and the safety concerns and potential conflicts of interest it’s generated along the way. It uses sources like corporate disclosures, legal complaints, open letters, media reports, and more — things that The Verge and other news outlets have frequently covered as individual stories, but gathered in one place here. Charts and data visualizations help illustrate what we know about OpenAI’s inner workings, including the nonprofits’ best guess at OpenAI’s corporate structure, its original capped-profit structure, and the proposed plan for its restructuring.
One prominent theme is how much OpenAI executives and board members stand to directly or indirectly gain from the company’s successes. The nonprofits include their best guess at CEO Sam Altman’s own investment portfolio, which includes a laundry list of companies — like Retro Biosciences, Helion Energy, Reddit, Stripe, Rewind AI, and Rain AI — that currently have overlapping business with OpenAI, such as partnerships, vendor relationships, or potential acquisition talks.
Johnston, who is executive director of The Midas Project, told The Verge that the project is partly about pointing out “the ways in which the vision that they had in the late 2010s departed from the way that they’re behaving today in 2025.”
The nonprofits behind the report said they received no funding, assistance, editorial direction, or “support of any kind” from Elon Musk, xAI, Anthropic, Meta, Google, Microsoft “or any other OpenAI competitor” and that the report has “complete editorial independence.” OpenAI declined to comment.
“We’re in an archival project here, where we’re showing what OpenAI was like then, what they’re like now… We’re just putting that information in front of the reader and asking the reader to draw their own conclusions about what to make of it,” Johnston told The Verge.
You can read the report at OpenAIFiles.org.
