The internet loves a good tech scandal, and “The OpenAI Files” seemed to deliver exactly that. A report packed with accusations against OpenAI and Sam Altman that had people asking: Is Sam Altman actually evil? But here’s the thing about sensational reports – they’re often long on dramatic claims and short on substance.
Tech YouTuber Theo did the heavy lifting that most people don’t bother with: he actually analyzed each claim in detail. His video “Is Sam Altman evil? The OpenAI Files are wild” systematically dismantles the report, and the results are pretty damning for the accusers. Most of the “bombshell” revelations turn out to be rumors, misunderstandings, or outright misrepresentations.
I covered this briefly in a Saturday post about three overblown news stories, but didn’t back up my claims about why these stories were fake. Time to fix that. Using Theo’s detailed analysis as the foundation, let’s go through each major accusation and see what actually holds water.
The Y Combinator Chairmanship: A Clerical Error Masquerading as Fraud
The OpenAI Files claim Sam Altman falsely listed himself as Y Combinator Chairman in SEC filings. Sounds serious, right? Except when you dig into the details, this appears to be a simple bureaucratic mixup during Altman’s transition from YC President to focusing full-time on OpenAI.
The reality is messier and more boring than fraud. During leadership transitions at organizations like YC, there’s often confusion about titles, interim roles, and who’s handling what responsibilities. Paul Graham himself clarified that this was likely a clerical error, not intentional misrepresentation.
More tellingly, YC and Altman maintain a good relationship today. Altman still appears on YC’s homepage, and there’s no bad blood between the organizations. If Altman had actually committed some kind of securities fraud involving YC, you’d expect the relationship to be… different. This claim appears to be a misunderstanding rather than intentional misrepresentation.
OpenAI’s Profit Cap Changes: Public Information, Not Secret Scandal
Another accusation involves OpenAI quietly changing its profit cap to increase returns by 20% annually without proper disclosure. This claim fundamentally misunderstands OpenAI’s corporate structure and treats normal business decisions as conspiracies.
OpenAI has a complex setup: a nonprofit that controls a capped-profit entity. As the company scaled and needed massive investment for AI development, shifting toward a more traditional for-profit model became inevitable. The competitive landscape demands it – you can’t build frontier AI models on nonprofit budgets.
The key point Theo makes is that these changes have been publicly known. OpenAI hasn’t been hiding their corporate structure shifts. When you’re competing with tech giants who can spend billions on AI development, maintaining arbitrary profit caps becomes a strategic liability, not a moral imperative. This change has been publicly known and is not a hidden scandal, as the report implies.
The Equity Stakes: Much Ado About Almost Nothing
The report breathlessly reveals that Altman held “indirect stakes” in OpenAI through Sequoia and YC funds, despite telling Congress he had no equity. This sounds damning until you realize what it actually means.
Theo’s comparison is perfect: this is like saying you have an undisclosed stake in Apple because you own shares in an S&P 500 index fund. Altman’s connection to OpenAI through YC is minuscule and spread across hundreds of other companies. Calling this a “significant personal equity stake” is misleading at best.
When Altman told Congress he had no equity in OpenAI, he was referring to direct ownership stakes, not fractional exposure through broad investment funds. The difference matters, and conflating the two is either ignorance or intentional misrepresentation. This claim of significant personal equity is largely insignificant and misleading.
Investment Partnerships: Standard Business Practice, Not Corruption
Another accusation involves Altman benefiting from OpenAI partnerships with companies he invested in, like Reddit and RainAI. This treats normal business networking as some kind of conspiracy.
Here’s how the tech world actually works: successful investors identify promising companies and invest in them. Later, when those companies prove valuable, they form partnerships with other successful companies. This isn’t corruption – it’s literally how business development happens.
The Reddit partnership is particularly amusing as criticism. OpenAI’s use of Reddit data for training has arguably hurt Reddit’s value more than helped it. If this was supposed to be some kind of insider dealing to benefit Reddit, it backfired spectacularly. Theo argues this is normal business practice; investing in promising companies and then forming partnerships is logical and not inherently collusive.
The Mythical 7% Stake: Old Rumors, No Evidence
One of the more persistent rumors was that Altman might receive a 7% stake in a restructured OpenAI, worth around $20 billion. The OpenAI Files treats this as fact, but it’s based on outdated speculation with inaccurate valuations.
As Theo points out, if Altman had received such a massive equity stake, it would have been disclosed publicly. Securities laws don’t allow you to hide billion-dollar ownership positions. The fact that no such disclosure exists should tell you everything about this rumor’s credibility. Theo notes this rumor is old and that the valuation mentioned was inaccurate, and there’s no evidence to support it.
Equity Clawback Provisions: Acknowledged, Apologized, Fixed
Here’s where the report actually found something real: OpenAI had problematic equity clawback provisions that could threaten departing employees who criticized the company. Altman initially denied knowing about these provisions despite signing documents authorizing them.
This was genuinely bad, but here’s the key difference from other claims: when confronted, Altman publicly acknowledged the mistake, apologized, confirmed the provisions were never enforced, and ensured they were fixed. This is how responsible leaders handle screwups – they own them and fix them.
Compare this to the typical corporate response of denial, deflection, and legal threats. Altman’s handling of this issue actually speaks well of his character, not poorly. Theo highlights that Altman publicly apologized, took ownership, stated the provisions were never enforced, and ensured they were fixed.
The Security Breach and Leopold: Much Less Than Meets the Eye
The report alleges a major security breach in 2023 that wasn’t disclosed for over a year, plus the firing of an employee named Leopold for raising security concerns. Both claims fall apart under scrutiny.
The “major breach” likely involved internal systems like Slack, not user data or core AI technology. Companies aren’t required to immediately disclose every internal security incident, especially when no customer data or proprietary technology is compromised. Theo suggests the breach likely involved internal messaging systems and didn’t compromise user data or core AI technology, thus not necessarily requiring immediate public disclosure.
As for Leopold, his subsequent 165-page manifesto on AI safety suggests someone with extreme views rather than a whistleblower revealing genuine misconduct. Sometimes people get fired because their approach to workplace concerns is unproductive, not because they’re truth-tellers. Theo also characterizes Leopold’s subsequent 165-page document on AI safety as extreme.
Ancient History and Irrelevant Drama
The OpenAI Files dredges up allegations about Altman’s behavior at his first startup, Looped, when he was 19 years old. Senior employees allegedly tried to get him fired for deceptive behavior.
This is ancient history – we’re talking about events from over 13 years ago when Altman was a teenager. People change and learn from mistakes, especially young founders navigating their first companies. Using decade-old workplace drama to attack someone’s current character is desperate. Theo dismisses this as irrelevant ancient history, noting that people change and young founders often make mistakes.
The Ilya and Mira Situation: Context Matters
The report claims OpenAI’s leading researcher Ilya Sutskever didn’t think Altman should “have the finger on the button for AGI,” and that former CTO Mira Murati expressed similar concerns.
But context matters. Sutskever has since defended Altman, apologized for his role in the attempted ousting, and stepped away from the board. Whatever concerns he had appear to have been resolved or reconsidered. Theo points out Sutskever’s unique personality and that Sutskever has since defended Altman, apologized for his role in the ousting attempt, and stepped away from the board.
Murati’s situation is even clearer – she has publicly refuted claims about opposing Altman and supported him during the board crisis. The report frames normal workplace feedback sessions as evidence of misconduct, which says more about the report’s agenda than Altman’s leadership. Theo presents Murati’s own statements where she refutes these claims, expresses support for Altman, and criticizes the old board’s actions. He also deconstructs a detailed account of Murati’s feedback to Altman, suggesting it was framed manipulatively and that HR involvement was a reasonable step for navigating internal feedback.
Anthropic’s Conflict of Interest
Perhaps the most revealing accusation comes from Anthropic’s founders, who described Altman’s management as “gaslighting and psychological abuse.” This sounds serious until you remember that Anthropic is OpenAI’s direct competitor.
Anthropic stands to gain significantly if OpenAI’s reputation is damaged. Taking their accusations at face value without acknowledging this massive conflict of interest is naive. Of course OpenAI’s competitors want to portray Altman as a terrible leader – it benefits their business. Theo highlights the clear conflict of interest, as Anthropic stands to gain significantly if OpenAI’s reputation is damaged.
Other Executive Feedback: Vague and Lacks Specifics
The claim that “at least five other OpenAI executives” gave similar negative feedback about Altman. Theo questions the source and vagueness of this, pointing out the small number of actual executives at OpenAI, making “at least five” a dubious claim without specifics.
The Startup Fund and Board Communications: Governance Issues, Not Scandals
Two final accusations involve Altman’s ownership of the OpenAI Startup Fund and allegedly misrepresenting board member opinions to other board members.
The startup fund ownership was a legitimate governance concern that was recognized and fixed by transferring ownership away from Altman. This is how healthy organizations handle conflicts of interest – they identify them and resolve them. Theo explains this was likely due to OpenAI’s nonprofit status at the time and Altman’s expertise in startup investing. He notes this was a legitimate concern, but the ownership has since been transferred to address these governance issues.
The board communication issue is classic “he said, she said” territory. The source appears biased, and saying different things to different people isn’t inherently deceptive – it depends entirely on context and intent. Theo argues saying different things to different people isn’t inherently deceitful and depends on context. He also criticizes the “he said, she said” nature of the claim and the source (New York Times, seen as biased).
Most accusations in the OpenAI Files fall apart under scrutiny, with percentages showing how debunked each claim appears.
Forcing Employees to Waive Whistleblower Compensation: Lacks Strong Evidence
Claim that OpenAI required employees to waive their federal right to whistleblower compensation, with former employees filing SEC complaints. Theo states he couldn’t find strong evidence for this, and that existing reports seem to contradict it or refer to the already-addressed overly restrictive non-disparagement clauses which Altman apologized for and fixed.
The Lobbying Accusation: You Can Support Regulation and Oppose Bad Laws
The final major accusation is that OpenAI publicly supported AI regulation while simultaneously lobbying to weaken the EU AI Act. This treats nuanced policy positions as hypocrisy.
It’s entirely possible to support regulation in principle while opposing specific legislation that’s flawed. The EU AI Act has significant problems – it’s vague, potentially economically damaging, and may not achieve its stated goals. Supporting better regulation doesn’t require supporting every regulation.
This is like criticizing someone for supporting healthcare reform while opposing a specific healthcare bill they think is badly written. Policy positions have nuance, and good-faith actors can disagree on implementation while agreeing on goals. Theo argues it’s possible to support regulation in principle while opposing specific bad legislation, and he believes the EU AI Act has significant flaws, such as vagueness and potential negative economic impact.
The Pattern: Rumors, Misunderstandings, and Agenda-Driven Reporting
When you step back and look at the full picture, a clear pattern emerges. Most of the “explosive” allegations in the OpenAI Files are either:
- Based on old rumors with no current evidence
- Misunderstandings of normal business practices
- Bureaucratic errors treated as intentional fraud
- Accusations from biased sources with conflicts of interest
- Vague claims lacking specific evidence
- Issues that were acknowledged and resolved
This doesn’t make Altman or OpenAI perfect. The equity clawback situation was genuinely problematic, and there are legitimate concerns about OpenAI’s corporate governance as they transition from nonprofit to for-profit operations. But those real issues get lost in the noise of manufactured outrage.
Why This Matters: The Cost of Fake Scandals
Fake scandals aren’t just annoying – they’re actively harmful. They distract from real issues, polarize discussions, and make it harder to have productive conversations about legitimate concerns.
OpenAI is building technology that could fundamentally change society. There are real questions about AI safety, corporate governance, competition policy, and the concentration of power in AI development. Those conversations are important and deserve serious treatment.
But when reports like the OpenAI Files flood the zone with weak accusations and conspiracy theories, they make it harder to focus on genuine issues. People become skeptical of all criticism, even legitimate concerns get dismissed, and the actual problems don’t get the attention they deserve.
This is why I spend time debunking obviously false narratives. It’s not because I think OpenAI is above criticism – it’s because bad-faith attacks make good-faith criticism less effective. When you consistently get the small stuff wrong, people stop trusting you on the big stuff.
The Broader Lesson: Critical Thinking in the Age of Information Overload
The OpenAI Files situation illustrates a broader problem with how we consume information about tech companies and their leaders. Sensational reports get massive attention because drama and conflict drive engagement. Measured analysis that finds “actually, most of this is overblown” gets much less viral spread.
This creates a systematic bias toward believing the worst about successful companies and leaders. The incentive structure rewards attention-grabbing accusations over careful fact-checking. And by the time thorough debunking happens, the original false narrative has already shaped public opinion.
The solution isn’t to automatically trust companies or dismiss all criticism. It’s to develop better critical thinking habits: check sources for conflicts of interest, look for specific evidence rather than vague accusations, consider whether dramatic claims align with other known facts, and respect the people doing the hard work of detailed fact-checking.
Theo’s video is a perfect example of what this looks like. He went through each claim methodically, provided context and evidence, acknowledged when criticism was legitimate, and came to measured conclusions. That’s the kind of analysis we need more of. This kind of nuanced analysis is crucial, especially when discussing topics like OpenAI’s strategic moves into new markets or the capabilities of new models like Google’s Gemini 2.5 Pro.
My Take: Focus on What Actually Matters
After going through all these accusations in detail, my conclusion is simple: most of the OpenAI Files is noise designed to generate attention rather than illuminate truth. The few legitimate concerns get buried under a pile of weak accusations and conspiracy theories.
This doesn’t mean OpenAI is beyond criticism. Their transition from nonprofit to for-profit raises real governance questions. Their competitive practices deserve scrutiny. Their approach to AI safety and transparency could be better. These are the conversations worth having. For instance, the discussion around the future of AI agents or the naming chaos of OpenAI’s GPT models are far more substantive.
But we can’t have productive conversations when the discourse is polluted by fake scandals and agenda-driven reporting. The OpenAI Files is a case study in how not to criticize powerful tech companies. It’s long on accusations, short on evidence, and ultimately counterproductive to the goal of meaningful accountability.
The lesson for anyone following AI developments is to be skeptical of sensational reports, especially those making dramatic claims without strong evidence. Look for detailed analysis like Theo’s that actually examines the facts.
The real issues in AI development are complex enough without adding manufactured drama to the mix.
