OpenAI, governance and the future of AI: the role of the Benefit Model

Full version of the article by Eric Ezechieli and Paolo Di Cesare, co-founders of NATIVA, published on Econopoly (Il Sole24Ore blog) on February 17, 2026.

When a company like OpenAI "sheds its skin," the market listens. This is not only due to the brand's weight or the speed with which its products have transformed the relationship between people and technology, but because every choice it makes—including those regarding governance—has the potential to become a benchmark. One of its latest moves, the transformation into a Public Benefit Corporation (known in Italy as a Società Benefit), should not be read merely as a legal adjustment. Rather, it is a signal that intercepts an increasingly strong trend within the tech sector toward Benefit models (which remain for-profit).

The company led by Sam Altman has reached a surreal valuation of $500 billion, a figure comparable to the GDP of all of Denmark. On one hand, it is true that OpenAI has lost its status as a non-profit organization—which it had been since its founding in 2015—and this choice has drawn sharp criticism from those who argue that innovations crucial to the future of humanity must be disseminated regardless of economic returns. On the other hand, it has chosen a new for-profit paradigm: that of a Benefit Corporation. This has allowed it to integrate a public interest mission into its Articles of Association and, consequently, its governance: “to ensure that artificial general intelligence (AGI)—highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity.”

OpenAI’s choice is not an isolated one. In recent years, a growing number of companies—tech and otherwise—have adopted Benefit models. These range from Anthropic to xAI (though the latter relinquished the status a few months ago), as well as entities like Patagonia, illycaffè, Chiesi Farmaceutici, Danone, Nespresso, Vodafone and Fastweb. Italy was the first sovereign country in the world to introduce Società Benefit in 2016; today, there are nearly 6,000, already representing 2% of companies with more than 250 employees.

The central point, often misunderstood, is that Benefit Corporations are not defined by the renunciation of profit, but by its redefinition. The corporate charter explicitly expands the scope of interests to be considered, including—alongside shareholders—people, communities, the environment and future generations. In this way, governance becomes the true "operating system" for evolving a company's impact, as shareholders grant management a broader mandate than the traditional one.

The Benefit model is a tool that enables a virtuous direction, but it does not make it automatic nor does it guarantee outcomes. Different Benefit Corporations can produce vastly different levels of economic, social, and environmental value, depending on how their purposes are translated and integrated into decision-making processes.

The OpenAI case represents a new frontier for experimentation. The new structure stipulates that the non-profit OpenAI Foundation maintains control over the Public Benefit Corporation and that the mission—identical for both Foundation and Corporation—is legally binding. However, substantial questions remain: how is the race for market share and profitability required by investors balanced in daily practice with the care needed when developing high-impact technologies? What decision-making mechanisms govern the company, and what systems of transparency and reporting allow stakeholders to assess whether the mission is truly being pursued? Furthermore, how does OpenAI intend to account for and manage the now-well-known negative social and environmental externalities resulting from the large-scale deployment of AI? As Andrew Kassoy, co-founder of B Lab, noted, an orientation toward the common good only has value if it becomes a priority decision-making criterion for governance and is accompanied by public, measurable, and reported commitments.

The stakes are incredibly high because artificial intelligence is not neutral: it reflects and amplifies the value, cultural, social, and economic paradigms of those who design it.

In this context, the Benefit model can represent a concrete direction for "channeling" the evolution of AI. Not as a mere label, and even less with the pretense of being "the solution," but as a governance infrastructure that makes it possible to hold together safety, inclusion, environmental impact, labor rights, and equitable access to technology. An AI industry built on a Benefit model effectively integrated into decision-making processes would have the chance to be truly oriented toward the common good. This should translate into product choices, measurement of effects with clear and standardized metrics, reduction of energy requirements and environmental impacts, the inclusion of currently underrepresented viewpoints, and support for the occupational transitions that automation inevitably generates.

This responsibility does not fall solely on those who develop the models. Companies adopting AI are also called to make decisive governance choices: when does AI truly strengthen the corporate mission, and when does it compromise it? What ethical limits cannot be crossed? And what minimum standards of transparency, security, energy impact, and bias management is it legitimate to demand?

Viewed from this perspective, the transformation of OpenAI—which must be held accountable by all stakeholders on how it honors what is written on paper and how it addresses the material ethical, environmental, and social issues regarding its products—becomes a testing ground for the entire system. The opportunity is to use this signal to build shared standards, clear policies, and credible reporting practices at the corporate and public policy levels globally. Benefit Corporations themselves can play a pioneering role here: translating principles into operational rules, from statutory purposes to binding policies on the use (and non-use) of AI, through to structured forms of reporting on social, environmental, and safety impacts.

The final question is simple and crucial: what legacy do we want artificial intelligence to leave? If the goal is widespread well-being for people and the regeneration of natural systems, a clear compass is indispensable—made of explicit purpose, transparent governance, and a plurality of perspectives—along with the courage to translate it into daily decisions. In this sense, Benefit Corporations are not a destination but a possible tool: a prototype that is beginning to demonstrate its potential and can contribute to steering a constructive dialogue toward a new way of conceiving business—one that is natively oriented toward regeneration.