MW OpenAI's biggest problem isn't AI safety. It's Sam Altman.
By Gustavo Razetti
'Conflict-averse' CEO has built a culture of groupthink - and a massive corporate liability
CEO Sam Altman has created a damaging "culture of silence" at OpenAI.
Every time a leader avoids a hard truth, sidesteps a real tension, or makes disagreement feel risky, you pay a hefty price.
The CEO of the most valuable AI company in history just confessed his real problem. And it has nothing to do with artificial intelligence.
In April, OpenAI's Sam Altman published a blog post responding to a New Yorker investigation. Somewhere between defending his character and processing a Molotov cocktail thrown at his home, he apologized for "being conflict-averse" - which has "caused great pain for me and OpenAI."
Great organizations are built on great conversations. And they collapse on the ones that never happen. Every time a leader avoids a hard truth, sidesteps a real tension or makes disagreement feel risky, you pay a hefty price: conversational debt.
At OpenAI, the bill came due publicly. Not once, but multiple times. Altman kept avoiding the same issue for too long until it became impossible to manage.
Avoiding difficult conversations is not a CEO personality quirk. It's a corporate liability. Here are five reasons why:
1. When disagreement becomes dangerous, groupthink takes over
Altman didn't build a culture of silence through cruelty. He built it through conflict-aversion. When a leader consistently sidesteps tension and doesn't model what it looks like to receive a hard truth, the team reads the signal: Disagreement is not safe here.
The result isn't debate. It's groupthink. People stop saying what they think and start saying what's expected. Helen Toner, a former OpenAI board member, testified that staff were "scared" to go against Altman. That's not a description of a tyrant. It's a description of a room where groupthink had replaced honest conversation - not through force, but through accumulated avoidance.
Make disagreement safe, not comfortable. You don't need a room full of people who agree with you. When dissent is welcome, problems get solved. When it isn't, they only get bigger.
2. Leaders must be open-minded, not just persuasive
When a leader wants to win instead of learn, people stop bringing real thinking to the room.
Altman has always been a charmer. Multiple insiders describe him as extraordinarily persuasive in one-on-one conversations. That's precisely the problem. Persuasion is good to drive buy-in. But dissent is critical to build better solutions, especially in the fast-paced, AI-driven world.
When a leader wants to win instead of learn, people stop bringing real thinking to the room. They manage how they present ideas, not what the ideas actually are. Conversations become performative, not productive.
After sharing your view, invite your team to challenge your thinking. Allow them to tell you what they think you're missing. Leaders who seek approval stop learning. Leaders who invite challenges make smarter decisions.
Read: Zuckerberg's 'Multi Mark' AI clone is a huge red flag for Meta shareholders
3. Unresolved disagreement doesn't disappear - it goes public
The board coup. Multiple experts' departures. The whistleblower SEC complaint. The NDA scandal. The New Yorker investigation. This pattern of avoidance at OpenAI escalated into a federal trial. Tesla $(TSLA)$ CEO Elon Musk, who co-founded OpenAI with Altman in 2015, sued him over the company's conversion to for-profit status. A disagreement that couldn't be resolved internally ended up in a courthouse. That's an expensive way to have a conversation.
Each crisis had its own news cycle and characters. But the structure was identical every time. A tension went unaddressed. People stopped raising it internally. And eventually it surfaced publicly: in a courtroom, on X, in a 16,000-word investigative report. Conversational debt works like financial debt. Deferring the payment doesn't reduce the balance. It compounds. At OpenAI, that interest was paid in governance collapse, talent loss and reputational damage.
Learn to read the early warning signs. If the same concern keeps surfacing in different forms, the conversation hasn't actually happened yet. Address it directly before the bill comes due.
4. The people who leave are the signal
Each high-profile deparature from OpenAI was part of a pattern.
Dario Amodei. Ilya Sutskever. Jan Leike, who posted publicly that safety culture had "taken a backseat to shiny products." Daniel Kokotajlo, who forfeited roughly $2 million in vested equity rather than sign a non-disparagement agreement and stay quiet.
Each of these high-profile deparatures from OpenAI was part of a pattern. All early signals were dismissed until people went silent. My own research shows that two-thirds of people stop speaking up when nothing changes. And the real cost is not just a headcount problem.
The exit interview is almost always too late. By the time someone resigns, the real conversation is already six months overdue. Ask your best people regularly what issues they've stopped raising and why.
5. When people can't disagree inside, they build your competition outside
Anthropic was founded in 2021 by Dario and Daniela Amodei, along with several senior former OpenAI colleagues. Their decision was brutal: Arguing with someone else's vision was unproductive. So they stopped arguing and started building.
The product they built - Claude, where values are made explicit, debatable and revisable - is a direct expression of a team that decided honest disagreement was non-negotiable. That principle isn't just a technical choice but a cultural statement. Anthropic's expected $900 billion valuation is the market's verdict on what OpenAI lost when it silenced that conversation.
Disagreement is a gift. It's cheaper, faster and far less damaging than when great ideas end up on your rival's desk.
Disagreement surfaced early is a gift. It's cheaper, faster and far less damaging than when great ideas end up on your rival's desk.
Teams don't rise to the level of their potential. They fall to the level of their conversations. Altman built one of the most powerful organizations in the world. He also made it harder for the people inside it to tell him the truth.
The most important question you can ask as a leader is not whether your team is aligned. It is whether your team has learned that speaking up is pointless. When disagreement dies, so does your business.
Gustavo Razzetti is CEO of Fearless Culture, a culture design consultancy. He is the author of "Remote, Not Distant, Stretch for Change," and "Forward Talk: The Bold New Method For Getting Teams Unstuck."
More: Elon Musk loses court battle with OpenAI's Sam Altman. These are the winners.
Also read: OpenAI might be filing to go public soon. How we got here.
-Gustavo Razetti
This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
May 27, 2026 07:50 ET (11:50 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments