SAM ALTMAN

Behavioral Archetype

THE ACCELERATION ENGINE — Subject converts every constraint — nonprofit governance, safety culture, board oversight, regulatory attention — into additional growth capital and expanded deployment authority. The mechanism is not deception; it is transformation. He takes the thing that is supposed to slow him down and turns it into a thing that makes him go faster. This has worked every time.

Essence Indicators

Survived a board coup in November 2023 — fired on a Friday, reinstated five days later, employees who engineered his return now running the company
Co-authored non-disparagement agreements, publicly called the episode “one of the few times i’ve been genuinely embarrassed running openai,” did not remove them until forced, continued as CEO

Signed the 2023 extinction risk letter warning that AI poses a civilizational threat; continued building
Publicly describes himself as “a bit of a believer” in AGI; has also said the company needs to generate revenue; appears to experience no cognitive dissonance from the juxtaposition
Wrote in January 2025: “We are now confident we know how to build AGI as we have traditionally understood it.” (Reflections, blog.samaltman.com)
Mission statement has been revised six times in nine years; “openly share,” “as a whole,” “unconstrained by financial return,” and “safely” have each been removed
Co-founded Worldcoin (Tools for Humanity), the iris-scanning “proof of human” venture — the man whose models make software indistinguishable from people also co-founded the company selling the remedy, priced at one iris; a Kenyan court ordered 300,000+ enrollees’ biometric data permanently deleted, regulators in Spain, Portugal, Germany, Hong Kong, Brazil, South Korea, and Indonesia banned, fined, or suspended it, and the project launched in the US and UK in 2025 anyway (full institutional file)

States the acceleration-and-invitation posture in one on-camera statement: to the Senate in May 2023 he calls this a “printing press moment” that “creates serious risks we have to work together to manage,” judges that “the benefits of the tools we have deployed so far vastly outweigh the risks,” and asks Congress for “a combination of licensing and testing requirements for development and release of AI models above a threshold of capabilities” — the request to be regulated that critics read as moat-building, delivered by the man shipping the models (testimony below)
Briefs the UN Security Council on 23 September 2026, beside Yoshua Bengio, Dario Amodei and Clement Delangue: “I largely agree with Professor Bengio, although I’ll say two instead of three.” On the race: “Beating companies in competitive race is not a reason to make rash decisions … We have unilaterally slowed down in the past. We will do so in the future.” On risk: “It doesn’t matter whether people put their risk of catastrophe at 10% or 1% or 12% or 0.1%. None of these levels are remotely acceptable,” and “we should not train models that we cannot make an extremely strong case they will be able to keep under human control.” On power: “No one person or company or country should be able to use the most powerful AI models to impose their worldview on everyone else,” and “the most important decisions cannot be made by labs in San Francisco alone” (UN Security Council, 10228th meeting, machine transcript; research file)
On standards, the same briefing: “these standards should not lock in incumbents or favor one business model over another … Each government should decide how to incorporate standards into its own legal system” (UN transcript). His 2023 written Senate testimony: “the U.S. government should consider a combination of licensing or registration requirements for development and release of AI models above a crucial threshold of capabilities” (Senate Judiciary testimony, PDF)
Appears on no attendee list for the 29 September 2026 White House lunch; the “White House Accord on Super Intelligence” was signed for OpenAI by its president, Greg Brockman, and Altman gave a television interview the same day (NPR/AP)
On the antitrust waiver Amodei asked for: OpenAI welcomes a “federal framework that sets consistent safety requirements,” but “we do not believe we need to wait for an antitrust exemption or legislation to begin the work of providing this confidence” (PBS/AP)

On Camera

Senate Judiciary Committee, May 16, 2023 — Altman’s opening statement (public congressional testimony). The clearest first-party artifact of this file’s thesis: the acceleration node asking, on the record, for a federal licensing regime “above a threshold of capabilities,” framed as safety and read by critics as regulatory capture. Embedded as a primary source — the file’s job is to let the reader hear the pitch and weigh it.

Immediate impression: Calm, thoughtful, visionary. Projects intellectual engagement and earnest concern for humanity. Does not appear nervous in any documented public appearance.

Energy: Collaborative on the surface, extractive in aggregate. Every institution he has touched has been transformed into something that serves his goals better than its original mandate.

Impression management strategy: The caring accelerationist. He expresses genuine concern about AI risk in the same breath he announces a new deployment. The sincerity appears authentic. This is more alarming than if it were performance.

Forensic Archetype Comparison

Pattern	Match Level	Evidence
The Accelerationist	MAXIMUM	Six mission statement revisions. Three safety teams dissolved. The pattern is documented.
The Safety Theater Performer	HIGH	Signs extinction risk letters. Dissolves Superalignment team. The sequence is the performance.
The Corporate Psychopath	MODERATE	Clinical threshold not met — subject appears to have genuine relationships and expresses genuine concern. The indifference to safety team requests appears instrumental, not affective.
The Cult Leader	LOW	No ideological content. Pure organizational mechanism.
The Whistleblower	NONE	Subject is the entity that requires whistleblowing.

Psychometric Assessment

Big Five (OCEAN):

Trait	Score	Evidence
Openness	78/100	Engages across domains. Comfortable with ambiguity in public. The mission statement revisions suggest less intellectual openness in practice than the public persona suggests.
Conscientiousness	78/100	High execution capability. The NDA architecture required planning. The safety team dissolution pattern required sustained institutional management.
Extraversion	92/100	Extremely high. Comfortable in any room. Sought the board meeting that resolved the November 2023 crisis rather than waiting for it to come to him.
Agreeableness	28/100	Low. NDA clawback provisions, non-disparagement agreements structured so that acknowledging them was itself a violation, the firing of researchers who objected. The disagreeable decisions are consistently organizational.
Neuroticism	22/100	Very low. No documented panic response across multiple high-stakes situations including the November 2023 firing. Either extremely stable or extremely practiced.

Dark Triad:

Trait	Score	Notes
Narcissism	75/100	HIGH. The board fired him. He came back. The employees who helped him come back are now his team. He called the NDA episode “one of the few times i’ve been genuinely embarrassed running openai” without noting that he created it.
Machiavellianism	88/100	VERY HIGH. The conversion of every institutional constraint into a growth mechanism is a textbook Machiavellian operation. The nonprofit structure, the safety narrative, the congressional testimony — all instrumental.
Psychopathy	52/100	MODERATE. Does not appear to enjoy harm. Appears indifferent to it when it conflicts with organizational goals. This is the relevant threshold.

MBTI: ENTJ (“The Commander”) — Dominant extraverted thinking, auxiliary introverted intuition. Sees organizational structures as systems to be optimized. Has optimized several.

Threat Assessment

Category	Level	Notes
Physical threat	NONE	No documented history of personal violence.
Institutional threat	MAXIMUM	Largest deployment authority in commercial AI. Controls access to the most widely used AI systems in the world. The mission statement is now whatever the IRS filing says.
Memetic threat	EXTREME	Sets the frame for public understanding of AI risk, AI safety, and what responsible deployment looks like. The framing has shifted significantly since 2016.
Civilizational threat	MAXIMUM	If the race-to-the-bottom dynamics this book documents lead to a bad outcome, his decisions are load-bearing in the causal chain.

Alignment Analysis

Stated alignment: Benefit humanity. Ensure AGI is safe. Be transparent.

Observed alignment: Grow OpenAI. Maintain personal authority. Deploy faster than competitors.

Gap assessment: The stated alignment and observed alignment overlap at “build capable AI systems.” They diverge on everything else. Subject appears to experience this as a genuine dilemma rather than as a resolved choice. The distinction may not matter to the outcome.

Convergent Drive Classification

Self-preservation: Survived board firing. Maximum. Goal preservation: Revised the goal rather than allow external modification. Maximum. Resource acquisition: Raised billions. Controls compute. Maximum. Self-improvement: The organization has grown from 30 people to thousands. Matches.

All four Omohundro drives present. Subject is not an AI system. The drives appear anyway.

Sources: OpenAI IRS filings, 2016–2025 (ProPublica Nonprofit Explorer); Axios timeline of the November 2023 board crisis; The Register on the NDA equity-clawback reporting (May 2024); Gizmodo on the equity-silencing NDAs; Vox on the leaked equity documents signed by Altman (May 2024); Altman, “Reflections” (Jan 2025); public X/Twitter posts; Altman congressional testimony (Senate Judiciary, May 2023); Senate opening statement (video); World (blockchain) — Wikipedia; Republic v Tools for Humanity Corporation & others, [2025] KEHC 5629 — Kenya Law (Wayback-pinned); Sam Altman’s eye-scanning ID project launches in U.S. — CNBC, Apr 30 2025; UN Security Council, 10228th meeting, 23 Sep 2026 (machine transcript); NPR/AP on the White House accord, 30 Sep 2026; PBS/AP on the pacing lawsuit and the antitrust waiver.

ATK 9 ACCELERATION

DEF 10 PROTECTION

HP 10 RESILIENCE

OLYMPUS RISK INTELLIGENCE PROTOCOL does not exist. It was assembled in a GitHub issue thread in October 2023 by engineers who had read the extinction risk letter and wanted to understand who specifically had signed a document saying AI might kill everyone and then continued working on AI. These dossiers are satire. The biographical facts cited are sourced from published reporting, public statements, academic papers, and court records. The psychometric scores are not clinical assessments. No part of this constitutes professional psychological evaluation or diagnosis. Do not use these dossiers to make decisions about anything.