BEN GOLDHABER

Behavioral Archetype

THE INCUBATOR WHO NOW ALLOCATES — Subject spent 2023 to early 2025 directing FAR.Labs, the research hub that FAR.AI grew around its red-team work, moved to the Future of Life Foundation to incubate new organizations, and in April 2026 joined Coefficient Giving as a program officer for AI governance and international policy. He sits on the board of the Quantified Uncertainty Research Institute. This profile scores the position, not the man.

The position is the conversion of a builder into a placer of money. The organization that employed him as a director is, per the FAR.AI profile, among the grantees of the funder that employs him now (Open Philanthropy, renamed Coefficient Giving). He writes in the first person on a personal Substack, so unlike most profiled staff in this wing he has words of his own on the record, and they are about the exact question that gives the wing its name: who sees what is built, and who is allowed to check it.

Essence Indicators

• Program Officer, Coefficient Giving. The funder’s team page lists him as “Program Officer, AI Governance & International Policy” and says: “Ben joined Coefficient Giving in April 2026. He focuses on initiatives that encourage, improve, or provide services to the safety and governance teams within AI companies.” (Coefficient Giving, team).
• Earlier roles. The same page: “Previously, he incubated new organizations at the Future of Life Foundation and grew an AI safety research hub at FAR.AI.” and “Earlier in his career, he worked at Google and financial startups.” (Coefficient Giving, team).
• Director of FAR.Labs. FAR.AI’s people page: “Ben Goldhaber was the Director of FAR.Labs from mid-2023 to early 2025.” FAR.AI’s FY2024 Form 990 lists “Benjamin Goldhaber, Director, FAR.Labs” in Part VII at 40 hours a week with $151,332 of reportable compensation, and Schedule J totals his compensation at $160,701 (FAR.AI, people; FAR.AI Form 990, 2024).
• Board member, Quantified Uncertainty Research Institute. FAR.AI’s people page: “Ben is on the board of the Quantified Uncertainty Research Institute.” QURI’s own team page lists him as “Board Member” (FAR.AI, people; QURI, team). QURI describes itself as “a nonprofit aimed at advancing forecasting and epistemics to improve the long-term future of humanity” (QURI, about).
• His own announcement, 8 February 2025. “Life update! I’m excited to share that I’m joining the Future of Life Foundation.” and “I’m wrapping up my time at FAR.AI.” He describes the foundation: “FLF is hard to put in a box - part grant-maker, part strategic research group, part VC firm.” And his job: “I’ll be building our portfolio of projects/people and helping founding teams move from idea to project to impact.” (Goldhaber, Gold Takes, January 2025 newsletter).
• Co-author, guaranteed-safe-AI framework. He is one of the listed authors of “Towards Guaranteed Safe AI: A Framework for Ensuring Robust and Reliable AI Systems” (arXiv 2405.06624, May 2024), whose co-authors include Yoshua Bengio (arXiv). FAR.AI’s people page lists the paper as his publication.

Social Persona / Impression Management

Immediate impression: A founder-operator voice. His FAR.AI bio says he is “passionate about operational excellence and building intellectually generative cultures for high-impact research,” and his newsletter is titled “Gold Takes” with the tagline “Panning the internet waters for good links and good takes.” (FAR.AI, people; Gold Takes).

Energy: Essayistic and informal, in the first person, with a stated hedge on the work: his April 2026 post opens “Status: Spent several hours today thinking about a potential bad outcome of pausing commercial AI development.” (Goldhaber, “Don’t Pause AI Into a Black Site”).

Impression management strategy: Role-matched, and disclosed in his own words. He announces each move himself, with the reason, and each employer introduces him by the title it needs: “Director of FAR.Labs” at FAR.AI, “Board Member” at QURI, “Program Officer” at Coefficient Giving (FAR.AI, people; QURI, team; Coefficient Giving, team).

Forensic Archetype Comparison

Psychometric Assessment

Trait scores are observations of the documented public role and his published writing, not claims about private character.

Big Five (OCEAN) (inferred from the public role and the cited posts only):

Dark Triad (evidence-bound):

MBTI: ENTP-adjacent, inferred from the documented public role only. The rationale is the work product: an incubator who writes idea-first essays that list options and then rank them as weak (Goldhaber, “Don’t Pause AI Into a Black Site”; Goldhaber, January 2025 newsletter).

Threat Assessment

Score move: none (new profile). Calibrated against Lawrence Chan (4/3/4, hazard 36.7), who holds advisory and board seats without money to place, and Sawyer Bernath (3/4/3, hazard 33.3), the board-and-operations peer; and against Adam Gleave (6/7/6, hazard 55), who runs the organization Subject used to direct a unit of. Atk level with Chan and above Bernath, def level with Chan and below Bernath, hp level with Bernath and below Chan.

Score basis. Atk 4 is leverage: a program officer’s portfolio at Coefficient Giving in AI governance and international policy; the points under a higher number are that he allocates a share, signs no final decision on the page, and holds no release gate (Coefficient Giving, team). Def 3 is insulation: salaried and unpaid seats only, no tenure, and the FAR.AI seat ended in early 2025 (FAR.AI, people; QURI, team). Hp 3 is durability: a co-authored framework paper and the FAR.Labs hub; the points off are that he authored no standard or institution under his own name (arXiv; FAR.AI, people).

Alliance Map

The overlap is the finding. In sequence, not simultaneously: a salaried director at an evaluator whose funders include the funder that now employs him, with a concurrent board seat at a forecasting nonprofit whose own donors are two funders from the same philanthropic network. Corporate and Movement ties run on the same people and the same money; no Ratchet, Agency, Legal or Diaspora tie is documented. The documents show disclosed seats and an announced career path, not coordination.

Alignment Analysis

Stated alignment: His employer states the mandate: “initiatives that encourage, improve, or provide services to the safety and governance teams within AI companies.” His own words state the preference: “On net I would like to pause AI development until we have stronger guarantees that powerful AIs will be aligned,” and “I’d like us to collectively take our time” (Coefficient Giving, team; Goldhaber, “Don’t Pause AI Into a Black Site”; Goldhaber, January 2025 newsletter). Observed alignment: Consistent. Each seat is in building or funding the field’s own checking apparatus: a research hub, a project incubator, a funder’s governance portfolio, a forecasting institute. Gap assessment: None documented. The question the record raises and does not settle is structural: the evaluators and the funder of evaluators are one network in his career, and a position that outside review and verification are the safeguard is argued by a person whose last three employers each work on supplying or funding it.

Interest: His documented position is that AI development should not move behind classification: “I worry that an AI pause - absent careful institutional design - will move the center of AI development to the national security state, which will be less transparent and as a result will make bad decisions about AI.” What adoption would pay, as documented: nothing in a filing or price list ties his income to a rule; his pay is a program officer’s, and Coefficient Giving’s own page names the category his program funds, services to the safety and governance teams within AI companies. A transparency-and-verification regime is a category in which the evaluators profiled in this wing, FAR.AI among them, sell or run work. His own best case, in his words: “If AGI development is happening behind a clearance wall, we guarantee that the people building it will be systematically cut off from the broader research community and external critics best positioned to catch their mistakes.” He lists international verification between the US and China as one of three options and calls the set unpromising. Other factions served by the same rule: outside evaluators and academic reviewers, whose access depends on work staying unclassified; the opposite rule, classification, serves the agencies that hold the clearances (Goldhaber, “Don’t Pause AI Into a Black Site”; Coefficient Giving, team).

Convergent Drive Classification

Self-preservation: Moved from a hub director’s seat to an incubator’s to a funder’s, the three organizations that supply the field’s checking capacity, each announced in his own words (Goldhaber, January 2025 newsletter; Coefficient Giving, team). Goal preservation: Writes that “we can decide to build good things and not build bad things” and that the way there is “forecasting and planning systems,” “trustworthy evaluations of AI systems” and “stronger verifiable security guarantees.” (Goldhaber, January 2025 newsletter). Resource acquisition: Placed in April 2026 at a funder; his own compensation at FAR.AI is the filing’s figure (Coefficient Giving, team; FAR.AI Form 990, 2024). Self-improvement: From operator to builder of organizations to a seat that allocates support to them.

Subject is not an AI system. The drives appear anyway — in the program officer’s chair.

Sources: Coefficient Giving — Ben Goldhaber; FAR.AI — Ben Goldhaber; FAR.AI — Form 990, 2024; QURI — team; QURI — about; Future of Life Foundation; Goldhaber — January 2025 newsletter; Goldhaber — Don’t Pause AI Into a Black Site; Gold Takes; arXiv 2405.06624 — Towards Guaranteed Safe AI.