ANNA MAKANJU

Behavioral Archetype

THE SITUATION ROOM TO THE SERVER ROOM — Subject ran national-security policy inside the United States government, moved to the platform that became the central object of the 2016-election inquiry to handle exactly that problem, and now runs global affairs for the frontier AI lab the next inquiry will be about. The throughline is not a single government or company. It is that the person managing how a frontier laboratory is perceived by states spent the formative chapter of her career inside the state, on its hardest national-security files, and the chapter after that managing a platform’s elections-and-policy exposure. She does not write the model’s refusals — and conflating her wing with the people who do is the error this file exists to avoid. She writes the terms on which governments decide whether the lab writing them should be constrained. The résumé is the exhibit: the Situation Room is upstream of the server room now, and the same person walked the hallway.

Essence Indicators

• Current OpenAI role: VP of Global Impact (since ~September 2024); previously VP of Global Affairs at OpenAI — the senior role positioning the company with governments worldwide

• U.S. government career: roles at the Department of Defense and State Department; Director for Russia on the National Security Council; and special adviser on Europe and Eurasia to then-Vice President Joe Biden

• Joined Facebook in 2018 to work on global elections and policy — the platform-elections file, at the company then at the center of the 2016 reckoning — before moving to OpenAI

• Credited in press with building OpenAI’s government relationships — the “political charm offensive” — making her the lab’s principal interface with the states that might regulate it
• The biographical fact the wing turns on: NSC Russia desk, then Facebook elections, then OpenAI global affairs. The path is the exhibit. The hand is not asserted on the model’s behavior — only on the lab’s politics.

Social Persona / Impression Management

Immediate impression: The seasoned diplomat. Fluent in the register of both the interagency process and the boardroom; the bearing of someone who has briefed principals and knows exactly how much to say in a hearing.

Energy: Relationship-first, institution-first. Does not argue a product decision. Builds the channel to the government and manages the lab’s standing inside it.

Impression management strategy: The public-interest convener. The framing — a frontier lab needs someone who understands national security and democratic governance at the highest level, and here she is — is correct, which is what makes it effective rather than suspect. The expertise is real. What the record adds is that the national-security judgment she brings is now deployed on behalf of a private laboratory’s relationship with the state that would regulate it.

Forensic Archetype Comparison

Psychometric Assessment

Big Five (OCEAN):

Dark Triad (held low and evidence-bound; the score measures structural position, not character):

MBTI: ENTJ-adjacent — strategic, organized around institutions and outcomes; sees the government relationship as a system to be built and steered.

Threat Assessment

Alignment Analysis

Stated alignment: Advance OpenAI’s mission in the public interest; help governments understand and govern AI; bring national-security seriousness to the frontier.

Observed alignment: Build and run OpenAI’s government relationships; position the lab advantageously with the states that would regulate it, from a national-security and platform-policy career.

Gap assessment: The stated and observed alignments overlap wherever “help governments govern AI well” coincides with “position OpenAI advantageously with those governments” — and the role is constructed so they tend to coincide. There is no documented gap between word and deed and no documented abuse of position; the hazard is structural and it is the Statecraft Wing’s defining one. When the person who once sat on the NSC’s Russia desk now manages a private laboratory’s relationship with the government, the public-interest framing and the corporate-interest framing travel together, and the record does not need to separate them for the arrangement to work. Note the wing boundary the file insists on: this governs the lab’s politics, not the model’s refusals. Conflating the two is the mistake that makes the larger thesis refutable. Her seat is statecraft, not model behavior.

Convergent Drive Classification

Self-preservation: Carries the statecraft method across every transition — NSC, State, DoD, Facebook, OpenAI. One skill, rising altitude. Goal preservation: Builds the government channel that frames how the lab is judged, so the terms of engagement are set before any single policy fight. Resource acquisition: Holds the scarcest resource in AI policy — trusted access to the governments that hold the regulatory pen. Self-improvement: Each move applies the same instrument at higher stakes: manage how power perceives the institution you serve.

Subject is not an AI system. The drives appear anyway — in the national-security hand that now governs a frontier lab’s politics.

Sources: Anna Makanju — Wikipedia; How Anna Makanju orchestrated OpenAI’s political charm offensive — Fast Company.