OpenAI, the maker of the most popular AI chatbots, said in its 2023 mission statement that it aims to build artificial intelligence that “safely benefits humanity, unconstrained by the need to generate financial gain.” But the creators of ChatGPT don’t seem to be as focused on doing this “securely” as they used to be.
While reviewing the latest IRS disclosure forms released in November 2025 and covering the year 2024, I noticed that OpenAI had removed changes such as “be safe” from its mission statement. This change in language coincided with the transformation from a nonprofit organization to an increasingly profit-oriented corporation.
OpenAI is currently facing several lawsuits over the safety of its products, making this change newsworthy. Many plaintiffs suing AI companies allege psychological manipulation, wrongful death, and assisted suicide, while others allege negligence.
As a scholar who studies nonprofit accountability and social enterprise governance, I see the removal of the word “safely” from their mission statements as an important change that has received little coverage outside of highly specialized media.
And I believe that OpenAI’s transformation is a test case for how we as a society oversee the activities of organizations that can bring great benefit but also catastrophic harm.
Tracing the origins of OpenAI
OpenAI, which also makes the Sora video artificial intelligence app, was founded in 2015 as a nonprofit scientific research organization. Its original purpose was not to make money, but to benefit society by publishing research results royalty-free.
To raise the funds needed to develop its AI models, OpenAI created a commercial subsidiary in 2019 under the leadership of CEO Sam Altman. Microsoft initially invested US$1 billion in the business. By 2024, the total amount will exceed $13 billion.
In return, Microsoft was promised a portion of future profits up to 100 times its initial investment. However, the software giant failed to win a seat on OpenAI’s nonprofit board. In other words, they lacked the ability to help steer the AI ventures they were funding.
A subsequent funding round in late 2024 raised $6.6 billion from multiple investors, but there was a catch. The funding would be debt unless OpenAI transforms into a more traditional for-profit business, where investors can own shares with no cap on profits and possibly get a seat on the board.
Establishment of a new system
In October 2025, OpenAI reached an agreement with the attorneys general of California and Delaware to become a more traditional for-profit company.
Under the new arrangement, OpenAI has been split into two entities: a nonprofit foundation and a for-profit company.
The OpenAI Foundation, a reorganized nonprofit organization, owns approximately one-quarter of the stock in OpenAI Group, a new for-profit public benefit corporation. Both are headquartered in California but incorporated in Delaware.
Public interest corporations are businesses that must consider interests beyond shareholders, such as society and the environment, and must issue an annual profit report to shareholders and the public. However, it is the responsibility of the board to decide how to weigh these benefits and what to report in terms of the benefits and damages caused by the company.
The new structure is outlined in a memorandum of understanding signed by OpenAI and the California Attorney General in October 2025 and approved by the Delaware Attorney General.
Many business media outlets reported on the move and predicted it would lead to further investment. Two months later, Japanese conglomerate SoftBank completed a $41 billion investment in OpenAI.
Mission statement changes
Most charities must annually file a form with the Internal Revenue Service detailing their mission, activities, and financial situation to demonstrate their tax-exempt status. The IRS publishes forms, giving nonprofits a way to communicate their mission to the world.
In its 2022 and 2023 forms, OpenAI said its mission is to “build artificial general intelligence (AI) that safely benefits humanity, unconstrained by the need to generate economic profit.”
IRS via Candid
That mission statement has changed as of OpenAI’s 2024 990 form, which the company filed with the IRS in late 2025. It now “ensures that artificial general intelligence benefits all humanity.”
IRS via Candid
OpenAI has removed its commitment to safety from its mission statement, as well as its commitment to making investors “unconstrained” by the need to make money. According to technology media Platformer, the company has also disbanded its “mission coordination” team.
In my view, these changes clearly demonstrate that OpenAI is prioritizing profits over product safety.
Indeed, OpenAI continues to mention safety when discussing its mission. “We see this mission as the most important challenge of our time,” the website says. “This requires simultaneous advances in AI capabilities, safety, and positive impact on the world.”
Review of legal governance system
A nonprofit board is responsible for making important decisions and upholding the organization’s mission.
Unlike private corporations, directors of tax-exempt charitable nonprofit organizations cannot personally enrich themselves by receiving a share of the profits. When a nonprofit owns a for-profit business, as OpenAI did in its previous structure, investors can receive a cut of the profits, but investors typically don’t have the opportunity to win a seat on the board or elect board members because it would be considered a conflict of interest.
The OpenAI Foundation currently owns 26% of OpenAI Group. This effectively means that the nonprofit board has relinquished nearly three-quarters of its control over the company. Software giant Microsoft owns a slightly larger 27% stake in OpenAI, having invested $13.8 billion in the AI company to date. OpenAI employees and other investors own the remaining shares.
Justin Sullivan/Getty Images
Seek further investment
The main purpose of OpenAI’s reorganization, dubbed its “recapitalization,” was to attract more private investment in the race for AI supremacy.
We have already succeeded in that respect.
As of early February 2026, the company is in talks with SoftBank for an additional $30 billion, and could receive up to $60 billion in total from Amazon, Nvidia, and Microsoft.
OpenAI’s valuation is now over $500 billion, up from $300 billion as of March 2025. The new structure also paves the way for an eventual initial public offering (IPO), which would not only allow the company to raise more capital through the stock market but also increase pressure for returns for shareholders.
OpenAI says the foundation’s endowment is worth about $130 billion.
OpenAI is a privately held company, so these numbers are just estimates. This means that these numbers are based on estimates of market value, rather than objective evidence such as market capitalization.
“We have secured concessions to ensure that charitable assets are used for their intended purposes,” California Attorney General Rob Bonta said in announcing the new structure. He also predicted that “safety will be the priority,” adding, “The number one priority is, and will continue to be, protecting our children.”
Measures that may help keep people safe
At the same time, several conditions in the OpenAI Reconstruction Note are designed to promote safety.
-
The Safety and Security Committee of the OpenAI Foundation Board of Directors has the power to “require mitigation actions,” which may include halting the release of new OpenAI products, based on its assessment of risk.
-
The for-profit OpenAI Group has its own board of directors and must only consider OpenAI’s mission when it comes to safety and security issues, not financial concerns.
-
The OpenAI Foundation’s nonprofit board may appoint all members of the OpenAI Group’s for-profit board.
But it would be difficult to hold the board accountable, given that neither the Foundation’s mission nor the OpenAI Group’s mission explicitly mentions safety.
Additionally, all but one of our board members currently serve on both boards, so we do not know how they will oversee themselves. And the memorandum signed by California’s attorney general does not indicate whether the attorney general was aware that references to safety had been removed from the mission statement.
Identify other paths OpenAI might take
There is an alternative model that I believe serves the public interest better than this.
When Health Net, a nonprofit health maintenance organization in California, converted to a for-profit insurance company in 1992, regulators required it to transfer 80 percent of its stock to another nonprofit health foundation. Unlike OpenAI, the Foundation had majority control after the transformation.
A coalition of California nonprofit organizations argued that the attorney general should require OpenAI to transfer all of its assets to an independent nonprofit organization.
Another example is the Philadelphia Inquirer. Pennsylvania Newspapers became a for-profit public benefit corporation in 2016 and is affiliated with the nonprofit Lenfest Institute.
This structure allows Philadelphia’s largest newspaper to attract investment without compromising its purpose of journalism that responds to the needs of the local community. This has become a model that could transform the local news industry.
At this point, I believe the people are bearing the burden of two governance failures. For one, OpenAI’s board has clearly abandoned its safety mission. And the other thing is that the attorneys general of California and Delaware have allowed it to happen.
