Headline & intro
Sam Altman’s public image used to be simple: the hyper-ambitious founder betting everything on AI. After a sprawling New Yorker profile and a scathing column in Ars Technica, that image now looks closer to "unelected infrastructure czar with a flexible relationship to the truth."
This is no longer just about one man’s ego. When the same personalities who built adtech and engagement-maximising social networks are put in charge of systems that will mediate knowledge, jobs, medicine and public discourse, the question becomes: do we trust them? In this piece, we’ll look beyond the headline drama and ask what this says about the entire AI industry — and why Europe, in particular, should be paying attention.
The news in brief
According to Ars Technica, a new 16,000‑word profile of Sam Altman in The New Yorker paints a disturbing picture of the OpenAI CEO and, by extension, of the culture driving the current AI boom. The profile, by Ronan Farrow and Andrew Marantz, draws on interviews with more than 100 people who have worked with or around Altman.
As summarised by Ars Technica, multiple sources describe Altman as willing to misrepresent facts, renegotiate or walk back agreements, and blur the line between ambition and reality. Former colleagues and board members reportedly question his honesty and even his basic trustworthiness, with some comparing the dynamic to other notorious Silicon Valley cases where hype outran reality.
The Ars piece also highlights Altman’s own techno-utopian writings, in which he imagines self-replicating robot supply chains and an almost frictionless transition to an AI-rich future, with little substantive engagement with social or political downsides.
Why this matters
The easy reaction is: "Silicon Valley founder turns out to be power-hungry and economical with the truth." That’s hardly news. What is new is the level of structural power now accumulating in the hands of a tiny number of such people.
OpenAI is not another photo-sharing app. Its models are quietly being integrated into office suites, government workflows, media production, education tools, and customer service infrastructure. When the CEO of such a company is described by close collaborators as willing to bend reality to his purposes, we’re dealing with a governance risk, not just a PR problem.
The core AI safety dilemma was always: how do you align powerful systems with human values? The Altman saga flips that question on its head: how do you align the humans who control those systems? If the people steering deployment are rewarded for racing ahead, downplaying externalities, and charming or bullying regulators, "AI alignment" collapses into a branding exercise.
There are clear winners and losers in the current setup. In the short term, founders and early investors benefit enormously from centralised control of models, data and compute. Large cloud providers, especially Microsoft in OpenAI’s case, gain lock‑in and a strategic moat against rivals. The losers are everyone who must treat these systems as neutral infrastructure: smaller companies building on opaque APIs, citizens whose data is scraped and monetised, and public institutions that risk dependency on vendors whose leadership they do not — and arguably should not — trust.
The immediate implication is a growing trust deficit. Criticism is no longer coming only from "AI doomers" or anti-tech activists; it is emerging from within Big Tech’s own orbit. That’s a signal regulators, enterprise buyers and voters should not ignore.
The bigger picture
Altman is not an outlier; he’s a particularly visible node in a pattern. Over the past decade we’ve seen a recurring archetype: the charismatic tech leader who presents themselves as a philosopher‑king of progress while treating laws, norms and sometimes basic honesty as optional.
Think of Elon Musk oscillating between launching rockets, undermining public transit, and turning X into a chaotic political loudhailer — all while pitching his own AI ventures as the safety‑conscious alternative. Or Mark Zuckerberg burning tens of billions on a metaverse pivot, then abruptly rebranding Meta as an "AI-first" company. Add Marc Andreessen’s manifesto declaring essentially that technology has no meaningful downside, and Peter Thiel’s mix of apocalyptic rhetoric and libertarian realpolitik.
Historically, we’ve seen this movie before: railroad barons in the 19th century, telecom and oil magnates in the 20th. Each time, optimistic narratives of progress coexisted with ruthless power plays. The difference now is speed and scope. AI systems can be deployed globally in weeks, and they intermediate not oil or steel but information, cognition and coordination.
OpenAI’s own governance drama in 2023 — a non-profit board briefly removing Altman, only to be reshaped under pressure from investors and staff — was an early warning. It showed how fragile "mission‑driven" structures become once tens of billions of dollars are on the table. The New Yorker revelations are a sequel, not a surprise.
Other AI labs are not immune. Anthropic, Google DeepMind, xAI and a long tail of model developers operate in the same incentive field: ship faster, promise more, dominate benchmarks, secure cloud subsidies. Some are more cautious than others, but the underlying business logic is similar. When that logic intersects with personalities optimised for fundraising and narrative control, we get exactly the overconfident, under‑accountable leadership style now under scrutiny.
The European / regional angle
For Europe, this is both a warning and an opening. On the one hand, European economies are becoming heavily dependent on AI services dominated by US (and increasingly Chinese) giants. When ministries, universities or SMEs plug into GPT‑style APIs, they are effectively outsourcing a slice of cognitive infrastructure to companies led by people like Altman and Musk.
On the other hand, the EU is building the most comprehensive regulatory framework for AI to date. The EU AI Act, combined with GDPR, the Digital Services Act and the Digital Markets Act, is explicitly designed to reduce the "trust me, bro" factor in digital infrastructure. High‑risk AI systems will require transparency, risk assessments and human oversight, regardless of how persuasive their CEOs are.
European startups such as Mistral AI in France or Aleph Alpha in Germany are trying to position themselves as more transparent, often more open, and closer to European public values. They are still dwarfed by US players in terms of compute and capital, but they operate under a regulatory and cultural environment that is more sceptical of founder worship.
For European users, the key question is not whether Altman personally is a villain or a visionary. It is whether critical public and economic functions should rely on black‑box systems controlled by a small circle of executives in San Francisco, subject mostly to US corporate law and venture incentives. The more troubling the picture of those individuals becomes, the stronger the case for European capacity — from open models to public compute — and for tough, enforceable guardrails.
Looking ahead
What happens next is unlikely to be a dramatic collapse; this is more a slow‑burn legitimacy crisis. OpenAI will continue to ship models. Microsoft will keep integrating them into Office, Windows and Azure. The New Yorker piece will fade from the feeds of most users.
But trust, once eroded, doesn’t easily return. Expect regulators in Brussels, national data protection authorities and competition agencies to look more closely at AI governance, not just AI outputs. Who sits on the board? Who has veto power? How are safety teams protected from commercial pressure? These questions, once niche, will move into the mainstream policy debate.
Enterprises and public institutions will start asking for more than glossy safety blogs. They will want audit rights, technical documentation, clear service‑level guarantees and credible red lines on data use. Procurement decisions — especially for education, healthcare and government — may become a powerful lever to favour vendors that can demonstrate institutional, not just personal, trustworthiness.
Technically, we’re likely to see continued momentum for open and locally deployable models. If, as Ars Technica’s columnist notes, many users would be happier with Wikipedia‑style governance and ethical training data, there is a market for less centralised, more democratically anchored AI. Europe’s public sector could play a catalytic role here by funding shared infrastructure and reference models.
The biggest unanswered question is cultural: will the next wave of AI leadership look more like infrastructure operators and less like messianic founders? If not, regulation will have to do the work that internal ethics failed to do.
The bottom line
The problem isn’t only that Sam Altman may be personally untrustworthy; it’s that our emerging AI infrastructure is structurally reliant on people optimised for hype, speed and power. For Europe and the wider world, the response cannot just be outrage tweets. It has to be a deliberate choice to build and reward different governance models — and to ask, before we integrate the next AI API: whose values, and whose incentives, are we wiring into our institutions?
