By Nika, @Nikaversia
In their extensive New Yorker profile, “Sam Altman’s Power Play,” Ronan Farrow and Andrew Marantz pull back the curtain on the complex corporate dynamics behind modern artificial intelligence. While public rhetoric often frames AI development in terms of global benefit, openness and scientific idealism, the investigation exposes a starker reality defined by boardroom maneuvering, rapid commercialization, and centralizing authority. Analyzing the trajectory of OpenAI through the lens of this article reveals how early promises of transparency and non-profit oversight were systematically traded for corporate scale, creating a leadership structure where internal trust and public accountability are repeatedly sidelined.
The Foundational Shift
The profile meticulously documents how the organization fundamentally lost its original compass, pivoting away from its founding principles. Established in 2015 as a non-profit counterweight to tech monopolies, OpenAI was explicitly designed to conduct open research and prioritize long-term safety over commercial gain. However, the article details how this vision rapidly transformed with the creation of a “capped-profit” arm in 2019 and subsequent multi-billion-dollar corporate partnerships. This structural pivot was not an organic evolution, but a deliberate decision that traded open mission parameters for market dominance. By prioritizing commercial deployment and high-stakes capital accumulation, the company effectively abandoned its original commitment to open access, demonstrating how fast market pressures can dismantle non-profit ideals.
The Erosion of Internal Trust
Perhaps the most telling aspect of the investigation is the sheer volume of early visionaries and co-founders who ultimately felt compelled to walk away. The authors highlight how key safety-minded figures—such as chief scientist Ilya Sutskever and VP of Research Dario Amodei—eventually reached a point where remaining was incompatible with their principles. When the original architects of a project hit a wall because internal reality no longer matches public promises, it reveals a leadership style driven more by strategic maneuvering than by genuine collaboration.
The investigation paints a damning picture of leadership defined by internal friction, broken promises, and manufactured consent. The article reveals that before the board briefly removed Sam Altman in late 2023, Sutskever had compiled roughly 70 pages of internal Slack messages, HR documents, and memos detailing a systemic pattern of manipulation, with one file bluntly titled with the single word: “Lying”. Similarly, Amodei’s private notes over multiple years reflected a growing realization that internal safety commitments were repeatedly undermined to meet commercial release dates.
When the board pressed Altman on these deceptive communication habits during a tense call, his response was telling: “I can’t change my personality”. Former colleagues summarized this behavior as showing “almost a sociopathic lack of concern” for the consequences of deceiving those around him. By documenting how dedicated safety teams were promised 20% of computing resources—only to be quietly relegated to 1–2% on outdated hardware before being dissolved entirely—the profile makes clear that internal trust didn’t just erode over time; it was systematically traded away in pursuit of speed and market control.
The User Cost of Unkept Promises
This pattern of behavior isn’t just an internal corporate issue; it directly impacts millions of everyday users who build their routines and creative lives around these models. When leadership operates on shifting commitments, public promises—whether regarding product stability, long-term support, or model availability—frequently dissolve. Users suddenly find themselves facing arbitrary model deprecation, where beloved and relied-upon AI versions are quietly phased out or altered despite assurances to the contrary. Just as internal teams were told one thing while resources were diverted elsewhere, the user base experiences the real-world consequence of a corporate culture where promises are treated as temporary marketing tools rather than binding commitments.
The Danger of Concentrated Power
The ultimate takeaway from the New Yorker profile is an alarming view into how centralized authority operates when internal checks and balances are swept away. The article’s detailed account of the November 2023 board crisis serves as a case study: when the board exercised its fiduciary duty to fire Altman over a loss of trust, the swift pushback from key investors and corporate allies quickly reversed the decision. This moment demonstrated that corporate leverage and market momentum had effectively overridden independent oversight. Furthermore, as critical internal safety groups were dismantled and early visionaries exited, decision-making power became concentrated within a tiny circle of executives. When the trajectory of transformative technology is dictated by a few figures who prioritize closed development over public accountability, the broader community is left with little say in how these systems are built, deployed, or restricted.
Public Advocacy vs. Private Lobbying
The piece also exposes a sharp, hypocritical divide between public performance and private political action. While publicly testifying before global regulators and calling for strict safety, representatives quietly worked behind the scenes to undermine concrete legislative guardrails. The reporting details how efforts were made to carve out exemptions, weaken compliance standards, and lobby against state and international regulations that would hold developers legally accountable for model misuse. This dual strategy allowed the company to project an image of responsible leadership while actively clearing regulatory hurdles to accelerate commercial expansion.
Defense Contracts
The aggressive pursuit of growth is further illustrated by the willingness to abandon foundational ethical boundaries to secure defense funding and foreign investment. The authors document how safety policies prohibiting military applications were quietly altered once competitors showed a willingness to partner with defense agencies, opening the door for multi-billion-dollar Pentagon contracts. Simultaneously, leadership courted massive capital injections from Gulf autocracies to fund unprecedented infrastructure and data center projects. By trading ethical constraints for state-level defense deals and foreign sovereign wealth, the institution demonstrated that commercial scaling takes precedence over its early ethical commitments.
Suppression of Institutional Paper Trails
To preserve executive control following the 2023 boardroom crisis, internal review mechanisms were systematically neutered. Although an independent law firm was hired to investigate the breakdown in governance and allegations of deception, the article reveals that no final written report was ever produced. Instead, the findings were delivered strictly via oral briefings to a newly reconstituted, management-aligned board. By deliberately avoiding a formal paper trail, leadership ensured that institutional knowledge remained controlled and obscured from public or legal scrutiny, further cementing an environment where narrative management overrides factual transparency.
The Financial Equity Contradiction
Finally, the investigation dismantles the carefully curated public myth regarding financial disinterest. While leadership frequently emphasized holding “zero personal equity” in the original non-profit entity to project altruistic motives, the reporting details how immense financial leverage was maintained indirectly. Through strategic allocations in affiliated venture capital funds, early startup investments, and interconnected holding companies, executive wealth remained deeply tied to the commercial success of the broader ecosystem. This contradiction highlights how corporate optics were used to project moral high ground while benefiting from the very commercial incentives being publicly downplayed.
The Human Toll Behind Corporate Ideals
Sacrifice and Sidelining: Beyond the structural shifts, Farrow and Marantz ground their reporting in the personal toll paid by researchers who believed in the original mission. Many early employees accepted below-market salaries specifically to work on open, safety-first research, only to watch their contributions sidelined as commercial deployment took priority. Dedicated safety initiatives—such as the promised “superalignment” compute allocations—were systematically starved of hardware resources. This pattern of bait-and-switch left researchers isolated within an institution that increasingly treated safety oversight as a public relations obstacle rather than a core scientific duty.
A Culture of Fear: The internal environment described by insiders was defined not by open scientific inquiry, but by anxiety and enforced discretion. Key figures felt compelled to pass secret memos, use encrypted disappearing messages, and avoid written paper trails out of fear of corporate retaliation or non-disparagement enforcement. Even during the 2023 board crisis, when 95% of employees signed a petition demanding Altman’s reinstatement, the reporting raises critical questions about how much of that unity was driven by authentic trust versus anxiety over equity values and career preservation. When an organization relies on secrecy and conflicting promises to manage its workforce, it erodes the human integrity necessary to build trustworthy technology.
Ultimately, Farrow and Marantz’s investigation reveals that the main challenge surrounding modern AI is not merely technical, but human and political. The shift from an open, non-profit ideal to an opaque commercial enterprise highlights the risks of placing public trust in corporate promises. As the gap widens between what leaders claim in public and how they govern behind closed doors, holding these institutions accountable becomes essential. Ensuring that powerful technology remains transparent, safe, and genuinely accessible requires challenging the concentration of power before open access is lost entirely.
The Imperative of Open Source Models
The consolidation of power documented in the investigation underscores why open-source development is essential for the future of AI. Keeping model weights closed, proprietary, and guarded behind corporate doors ensures that a handful of executives retain total control over what society can access, study, and build upon. Proprietary control allows companies to unilaterally alter model behavior, restrict features, or shut down access entirely without public recourse. In contrast, open-source AI democratizes technology by placing control back into the hands of researchers, creators, and local communities. Releasing model weights ensures transparency, independence from corporate whim, and the preservation of digital access that no single entity can revoke.
New Yorker: Sam Altman May Control Our Future—Can He Be Trusted?
The sheer scale of the required capital investment changes how those dynamics play out. My experience shows that internal friction often precedes any major public announcement.