The Midas Project

Background

The Midas Project is an artificial intelligence (AI) safety-advocacy nonprofit founded in 2024 by Tyler Johnston. The group advocates for stricter safety standards and increased transparency at companies that build and operate AI platforms, including OpenAI, Meta, Google, and Microsoft. ^{1 2}

Activities

In January 2026, the Midas Project launched Model Republic, an online publication that covers AI public policy. The outlet’s first article criticized the AI advocacy efforts of the venture-capital firm Andreessen Horowitz. Model Republic’s second article reported on an alleged “coordinated” attempt by right-wing pundits to oppose the AI OVERWATCH Act. ³

In August 2025, the Midas Project launched an open letter with a list of seven numbered requests for OpenAI to maintain transparency after the organization restructured from a nonprofit entity into a nonprofit that governs a for-profit public-benefit corporation. The letter expressed concerns that under its new structure, OpenAI would operate as a for-profit entity with fewer legal requirements for transparency, including regarding its commercialization of AI products and equity distribution. Signatories of the open letter include “more than 100 prominent AI experts, former OpenAI team members, public figures, and civil society groups,” including Ethereum co-founder Vitalik Buterin, Vox cofounder and left-wing pundit Matthew Yglesias, MIT physicist Max Tegmark, University of Texas-Austin computer scientist Scott Aaronson, and British actor Stephen Fry. ^{4  5}

In July 2025, the Midas Project published a critique of Anthropic’s AI safety framework, claiming that the guidelines’ “risk thresholds are extraordinarily high.” ⁶

In June 2025, the Midas Project published the OpenAI Files, an investigation into OpenAI’s “governance practices, leadership integrity, and organizational culture.” The report was highly critical of OpenAI, including its removal on profit caps, the allegedly domineering conduct of CEO Sam Altman, forcing employees to sign “highly restrictive” non-disclosure agreements, and cultivating what the New York Times characterized as “a culture of recklessness and secrecy.” ⁷

In February 2025, the Midas Project launched the Seoul Commitment Tracker to monitor progress on AI safety commitments made by 16 leading AI companies at the 2024 AI Safety Summit hosted in Seoul, South Korea. As of February 2026, the Tracker had claimed that six companies had fulfilled their promises (including Amazon, Google, Microsoft, and OpenAI), four companies have partially fulfilled their commitments (including Meta), and six companies have not fulfilled their commitments (including IBM and Samsung). ⁸

Conflict with OpenAI

The Midas Project has been highly critical of OpenAI, having released the OpenAI Files and issuing an open letter calling for more transparency at the company. OpenAI has responded with attacks against the Midas Project that have been characterized as a “campaign against critics.” ²

In September 2025, OpenAI issued a subpoena to the Midas Project and Midas Project executive director Tyler Johnston as part of the ongoing legal conflict between OpenAI and Elon Musk— who co-founded OpenAI, but left in 2018. In the subpoena, OpenAI alleged that the Midas Project (and other similar nonprofits) were established and funded by Musk and possibly Meta CEO Mark Zuckerberg as a covert means of publicly attacking OpenAI. The subpoena demanded that the Midas Project disclose any communication with Musk and Zuckerberg and a list of its donors. The Midas Project and Johnston have denied ever receiving funding from Musk or Meta or speaking to Musk about the organization. ²

In January 2026, in an interview, Johnston accused OpenAI of running a smear campaign and trying to silence the Midas Project with fear tactics: “It’s a bit scary to know that the most valuable private company in the world has your address and has shown up and has questions for you.” Johnston claimed that the negative coverage of the Midas Project caused insurance brokers to refuse to cover his nonprofit. ⁹

Leadership

Tyler Johnston is the founder and, as of 2025, was executive director of the Midas Project. In September 2025, Johnston claimed he was the only full-time employee of the organization and earned a salary of about $25,000 per year. ^{2  10}

From May 2022 to August 2023, Johnston worked as a corporate-communications specialist at the Humane League, an animal-rights nonprofit. From May 2021 to October 2022, he worked as an editorial assistant at the Kidney Foundation, and in mid-2021, he was a corporate-engagement research fellow at the Good Food Institute. ¹⁰

Prior to founding the Midas Project, Johnston received a $35,000 grant from Open Philanthropy to help transition his career to AI policy. ²

As of 2025, Johnston was on the board of PauseAI, another AI safety group. ²

Financials

As of February 2026, the Midas Project had not filed publicly disclosed tax forms other than the IRS e-Postcard declaring gross receipts of less than $50,000 in 2024. ¹¹ In 2024, the year of its founding, the Survival and Flourishing Fund recommended the Midas Project for a $31,000 grant. ¹²