DUSTIN MOSKOVITZ

Behavioral Archetype

THE PRINCIPAL — Subject is the person whose personal fortune is the capital behind the largest funder of the AI evaluator and AI safety field. He does not choose the grants. Coefficient Giving (formerly Open Philanthropy) does that, as the outsourced staff of the foundation he runs with his wife. He is the principal in the principal-agent line: the Facebook fortune became Asana, Asana became a controlling stake, and the proceeds became a foundation whose adviser is the grantmaker that paid for Redwood Research, Epoch AI, FAR.AI, Apollo Research and most of the rest of the evaluator layer. He is also, by his own account, a board observer at Anthropic, whose models the same evaluators grade. This profile scores the position, not the man.

Essence Indicators

• Founder-donor of Good Ventures. Good Ventures describes itself as “a philanthropic foundation created by Cari Tuna and Dustin Moskovitz,” whose “primary philanthropic advisor” is Coefficient Giving. The founders’ letter says “When we founded Good Ventures in 2011, we knew we wanted to give away the vast majority of our wealth,” and that “With support from the Coefficient team, Good Ventures has distributed more than $7 billion.” The foundation’s governance page lists the board as Cari Tuna (Chair), Dustin Moskovitz (Secretary and Treasurer) and Divesh Makan of ICONIQ Capital. (Good Ventures, About; Good Ventures, Our Story; Good Ventures, Governance)

• Main funder of Coefficient Giving. Coefficient’s own team page for him says “he and Cari are Coefficient Giving’s main funders,” lists him under “Board Member,” and says he “also previously co-founded Facebook.” Its rename announcement calls Good Ventures “the foundation of Cari Tuna and Dustin Moskovitz,” says “Cari and Dustin have committed to spending down the vast majority of their wealth within their lifetimes,” and says “We have helped them give over $4 billion so far.” (Coefficient Giving, Dustin Moskovitz; Coefficient Giving, Open Philanthropy is now Coefficient Giving)

• The scale of the AI line, as Coefficient states it. “In 2024, Coefficient Giving committed $168 million across our technical AI safety, security, and fieldbuilding work,” and “we more than doubled that to $351 million, and this year we’re on track to commit over $1 billion.” The same post says it “recently approved $160 million for” Resolution and “recommended a $68 million renewal over 3 years” for Epoch. (Coefficient Giving, We’re urgently scaling our work on AI and biosecurity)

• Where the money lands, from the grant listings. Coefficient’s grant pages record, among others: Redwood Research $5,300,000 for general support (Sep 2023); FAR.AI $28,675,000 over three years (Oct 2025); Apollo Research $2,178,700 for general support (Jun 2024); Epoch $6,922,565 for general support (Jun 2023). The full table of recipients and amounts, with the Survival and Flourishing Fund lines beside it, is under “Shared money” on the evaluator alliances map. (Redwood; FAR.AI; Apollo; Epoch)

• Named Funder of the Survival and Flourishing Fund’s 2026 round. SFF ran its single 2026 round “for Funders Jaan Tallinn and Dustin Moskovitz,” recommending $63.13MM in total, of which $29.03MM is “Recommended to be funded by Dustin Moskovitz” ($26.79MM with distribution pending, $2.25MM covered by Speculation Grants funded by Jaan Tallinn). The Moskovitz lines on evaluators: Epoch AI $542,000 of $646,000; FAR.AI $771,000 of $875,000; Apollo Research $306,000 of $315,000; Palisade Research $1,360,000 of $1,842,000; SecureBio $896,000 of $927,000. METR is listed at $2,201,000 with funders “Jaan Tallinn & Dustin Moskovitz” and no per-funder split printed. Redwood Research does not appear in the round. (Survival and Flourishing Fund, SFF-2026 Main Round Recommendations)

• Anthropic, as investor and observer. Anthropic’s May 2021 Series A announcement says “The round included participation from James McClave, Dustin Moskovitz” and others. Asana’s 2026 proxy statement says “Dustin Moskovitz, Chairman of our Board of Directors and our former Chief Executive Officer, has a minority ownership interest in Anthropic,” and that Asana incurred $1.1 million in AI-services expenses with Anthropic in its fiscal year to 31 January 2026. In his own words, October 2025: “I’m a board observer at Anthropic and we are early donors to OpenAI.” (Anthropic, Anthropic raises $124 million; Asana, 2026 Proxy Statement; Stratechery, An Interview with Asana Founder Dustin Moskovitz about AI, SaaS, and Safety, 20 Oct 2025)

• His account of the investment, as POLITICO reported it. POLITICO quotes him saying returns from the Anthropic investment “will be entirely redirected back into our philanthropic work.” It also reports that “In 2017, Moskovitz’s Open Philanthropy provided a $30 million grant to OpenAI,” which Open Philanthropy’s spokesperson described as support for AI safety work and not an equity investment. (POLITICO, How a billionaire-backed network of AI advisers took over Washington, 13 Oct 2023)

• Asana: founder, former CEO, controlling holder, current director. Asana’s proxy statement says his last day as President and CEO was 21 July 2025 and that he “transitioned to serving solely as a non-employee director and the Chair of our Board.” Its 8-K of 25 September 2026 says the board appointed Dan Rogers Chair to succeed him and that “Mr. Moskovitz will continue to serve on the Board as a Class I director.” The proxy’s ownership table lists him with 74.2 percent of total voting power (Class B carries 10 votes per share), and its footnotes say his holdings include 1,720,916 Class A shares held of record by Good Ventures Foundation, on whose board he and his spouse sit; 19,273,127 shares bought in a 2022 private placement carry an irrevocable proxy to Asana, so that “Mr. Moskovitz does not retain voting power” over the Voting Neutralization Shares. (Asana, 2026 Proxy Statement; Asana, Form 8-K, 25 Sep 2026)

• In his own words on AI risk (Medium, February 2024). “At the beginning, I thought I knew AI would unfold safely and even helped to build it faster,” and “concluded that the risks were big enough and real enough to be taken much more seriously than humanity currently does.” In a footnote: “I would even say I started off as somewhat of an AI accelerationist in the early 2010s as an investor in Vicarious.” On the confident: “the people I least understand in the AI risk debate are the ones who have ~100% confidence that AI will or will not destroy us.” On his own motive: “my interest in mitigating the potential risks isn’t rooted in specific predictions about the far future, but rather simple concern about the next few decades.” On the grantmaker: “they’ve also made grants over the years to folks who are a less perfect match.” (Dustin Moskovitz, Works in Progress: The Long Journey to Doing Good Better, linked from Good Ventures, Speaking and Writing)

• Stratechery interview, October 2025. “I think these technologies are really dangerous, I think we’re going to have something pretty bad happen”; “you should move a little slower in order to move fast”; “we work on AI safety and those things meet in a very dangerous world”; on state rules, “the whole patchwork of regulations”; on the pause letter, “I’m not on that letter.” (Stratechery, 20 Oct 2025)

• CNBC, June 2023. “I definitely think there’s a big risk there,” and on the rule he wants: “making sure that state-of-the-art later generations, like GPT-5, GPT-6, get run through safety evaluations before being released into the world,” adding “I think that will require regulation to coordinate all the players.” CNBC reports that he invested in Anthropic in 2021. (CNBC, Asana’s Dustin Moskovitz is bullish on AI but concerned about risks, 24 Jun 2023)

• Giving Pledge, 2010. The Giving Pledge lists Dustin Moskovitz and Cari Tuna as signers in 2010; his letter says “I view that reward not as personal wealth, but as a tool.” (The Giving Pledge)

• Social accounts: none verified. Good Ventures’ writing page links his Medium essay, and no page he or Good Ventures controls was found linking a personal X account. No handle is added.

Social Persona / Impression Management

Immediate impression: The reluctant billionaire-engineer. Co-founder of the most visible social network, then of a work-management company, then of a foundation that spends the proceeds down. The public posture is spreadsheet-first: prioritisation, uncertainty and “humility” rather than doctrine.

Energy: Principal, not operator. Delegates grant selection to the grantmaker he funds, delegates Asana to a successor CEO, and writes at length about method. Appears in the record mostly through filings, a founders’ letter, one long essay and a handful of interviews.

Impression management strategy: The hedger of his own movement. His essay distances him from both camps, the confident doomers (“near 100% probability of doom, like Eliezer Yudkowsky”) and the confident dismissers, and from the effective-altruism label as a creed. He describes the grantmaker’s grantees as sometimes “a less perfect match.” First-hand exhibit: his Pledge letter frames the fortune as “a tool,” and the Medium essay is the one document in which he states his AI view at length (The Giving Pledge; Medium).

Forensic Archetype Comparison

Psychometric Assessment

Big Five (OCEAN):

Dark Triad:

MBTI: INTP (“The Logician”): dominant introverted thinking, auxiliary extraverted intuition. The essay treats philanthropy as a standing question under uncertainty, and the record shows a preference for frameworks over assertions. Observation of the public role only.

Threat Assessment

Score basis. Atk 8, def 10, hp 9 at hazard 88. Atk is leverage by funding allocation: Coefficient’s own post puts its AI commitments on track for over $1 billion in 2026 and SFF-2026 recommends $29.03MM as funded by him (Coefficient; SFF-2026). It is one point above Jaan Tallinn because the scale of the allocation is an order of magnitude larger (a $63.13MM SFF round against a grantmaker committing hundreds of millions a year), and it is not higher because grant selection sits with the grantmaker’s staff and he holds no deployment or policy authorship. Def is insulation: Asana’s proxy documents 74.2 percent of voting power, and Good Ventures is his and Cari Tuna’s own foundation (Asana). Asana’s purchases from Anthropic ($1.1 million) and OpenAI ($0.3 million) are the only payments in the record from the graded set to a company he controls, and they are small against the holding. Hp is durability: Coefficient is now a multi-donor institution with multi-year commitments, offset by the couple’s stated spend-down (Coefficient). Peers on the same values: Eric Schmidt holds 8/10/9 at hazard 88: the same self-funded seat and the same early Anthropic stake, with operating lines (a drone contractor, a rocket company, a national commission) that Moskovitz lacks, set against Moskovitz’s larger allocation through a standing grantmaker; the two net to the same value. Holden Karnofsky holds 8/9/8 at hazard 82: he allocated the grants and now writes policy inside a lab, but the household money behind the institution is not his; the 1-point gaps on def and hp are that difference. Open Philanthropy holds 8/8/9 at hazard 80 as the organisation; the principal sits above it on insulation.

Alignment Analysis

Stated alignment: Mitigate the risks from advanced AI while keeping its benefits. In his words: “I believe we can do a lot to mitigate them and still get the benefits of transformative AI, but only if we try” (Medium). Give the fortune away within his lifetime.

Observed alignment: Fund, through one grantmaker and one fund, most of the institutions that define what “safe enough to ship” means, while holding equity in and a board-observer seat at one of the labs they grade, and running a public company that buys from two of them.

Gap assessment: The stated and observed alignments overlap wherever “fund AI-risk mitigation” coincides with “fund the evaluators of a lab you also invested in.” He states the overlap himself: returns on the investment “will be entirely redirected back into our philanthropic work” (POLITICO) and “we work on AI safety and those things meet in a very dangerous world” (Stratechery). The record documents the structure. It does not settle motive, and for the principal it never needs to.

Interest: What each position would pay him or his holdings, documented, with his best case.

• Pre-release safety evaluation, backed by regulation. Position, in his own words: “making sure that state-of-the-art later generations, like GPT-5, GPT-6, get run through safety evaluations before being released into the world,” which “will require regulation to coordinate all the players” (CNBC). What it pays: a mandatory-evaluation rule widens the market for the evaluators the grantmaker he funds supports (Redwood, Epoch, FAR.AI, Apollo and SecureBio appear on its or SFF’s lists above), and he holds a minority interest in Anthropic (Asana proxy). The record documents no Anthropic holding figure. His best case: “I think these technologies are really dangerous, I think we’re going to have something pretty bad happen” (Stratechery). Who else gains: the labs that want one predictable gate, and the government institutes given the evaluation function.
• One rule instead of a patchwork. Position: “the whole patchwork of regulations” is his phrase for state-by-state rules, and he asks for one set first (Stratechery). What it pays: Asana buys AI services from Anthropic ($1.1 million) and OpenAI ($0.3 million) and sells subscriptions to both, and he holds Anthropic equity (Asana proxy); a single federal rule lowers compliance cost for every firm in that chain. His best case: “you should move a little slower in order to move fast.” Who else gains: every company subject to a state regime, which is the faction the phrase describes.
• Funding AI-risk mitigation at scale. Position: Coefficient commits “over $1 billion” on its AI line in 2026 (Coefficient). What it pays: the grantees on its listing and SFF’s table, which include the evaluators the labs use, and, by his statement, returns from his Anthropic investment go back to philanthropy. His best case: his interest “isn’t rooted in specific predictions about the far future, but rather simple concern about the next few decades.” Who else gains: other donors, since Coefficient says “Open Philanthropy and Good Ventures are currently a concentrated share of AI safety philanthropic funding” and it advises others (Coefficient).

Interest is a documented fact about each position, not a claim about motive, and not a verdict that the argument is wrong. The best case is stated at the strength he states it.

Alliance Map

Several of these run at the same time. In 2026 he is the principal of the foundation whose adviser funds the evaluators, a named Funder of the fund that funds the same evaluators, an investor in and board observer at the lab those evaluators grade, and the controlling holder of a company that buys from that lab and from OpenAI. The overlap is the finding, not a coordination claim.

Convergent Drive Classification

Self-preservation: Survives every transition by carrying the capital, not the title. Founder, CEO, Chair, director, pledger: the money stays and the seat changes. Goal preservation: Funds the institutions that define the goal, through an intermediary, so the goal is protected by the grant round before it is debated. Resource acquisition: Converts one fortune into a standing grantmaking institution that now also advises other donors, which is the scarcest resource in the apparatus: money that arrives every year. Self-improvement: Each stage reuses the same instrument at a larger scale: Good Ventures, then Coefficient’s rename and pooled funds, then a named seat on SFF’s round.

Subject is not an AI system. The drives appear anyway — in the principal whose product is the continued existence of the institutions that decide what “safe” means.

Funding record: The grantmaker’s place in the safety-funding network is on The EA to AI-Safety Funding Pipeline, and the recipients’ shared money is on the evaluator alliances map.

Sources: Good Ventures, About; Good Ventures, Our Story; Good Ventures, Governance; Good Ventures, Speaking and Writing; Coefficient Giving, Dustin Moskovitz; Coefficient Giving, Open Philanthropy is now Coefficient Giving; Coefficient Giving, We’re urgently scaling our work on AI and biosecurity; Coefficient Giving, AI safety and security need more funders; SFF-2026 recommendations; Anthropic, Series A; Asana 2026 proxy statement; Asana Form 8-K, 25 Sep 2026; Dustin Moskovitz, Medium essay; Stratechery interview; CNBC, 24 Jun 2023; POLITICO, 13 Oct 2023; The Giving Pledge; Dustin Moskovitz, Wikipedia (index).