CARI TUNA

Behavioral Archetype

THE CHAIR OF THE CHECKBOOK — Subject is the co-founder of the foundation whose adviser pays for much of the AI evaluator layer, and the chair of the boards of that adviser’s grantmaking and research entities. She does not choose the grants; she says so. She is the other principal in the principal-agent line that Dustin Moskovitz heads: the foundation’s Form 990-PF lists her as President and Board Chair, and Coefficient Giving, formerly Open Philanthropy, lists her as Chair of the Board of Managers of the entity whose staff recommend grants. Her own page at Coefficient says she “serves as the Chair of Coefficient Giving” and “also serves on the board of GiveWell.” In April 2026 she described what the AI line buys: “funding evaluations of potentially dangerous capabilities like cyberoffense or biological research.” This profile scores the position, not the woman.

Essence Indicators

• Co-founder and Chair, Good Ventures. Coefficient’s page on her: “Cari Tuna is the co-founder and Chair of Good Ventures, a philanthropic foundation she created with her husband Dustin Moskovitz in 2011. Since then, Cari and Dustin have given away more than $4 billion to high-impact causes. They were among the first donors to sign the Giving Pledge, when Cari was just 25 — making her the youngest signatory in the Pledge’s history.” It adds that she “was previously a reporter for The Wall Street Journal and has a B.A. in Political Science from Yale University.” Good Ventures’ governance page lists Cari Tuna as “Chair,” Dustin Moskovitz as “Secretary and Treasurer” and “Divesh Makan, Partner, ICONIQ Capital” (Coefficient Giving, Cari Tuna; Good Ventures, Governance). See Good Ventures.

• The foundation’s return. Good Ventures Foundation’s Form 990-PF for the tax year 1 July 2024 to 30 June 2025 lists “CARI TUNA” as “PRESIDENT, BOARD CHAIR, DIRECTOR,” one hour a week, $0, with Dustin Moskovitz as “SECRETARY, TREASURER, DIRECTOR” and Divesh Makan as “DIRECTOR.” It reports $306,706,819 in contributions, gifts and grants paid and $10,107,955,038 in assets. Its Part XV answers the question of which managers contributed more than 2 percent of total contributions with “CARI TUNA” and “DUSTIN MOSKOVITZ.” Schedule B, Part I, lists two contributors: “Dustin A Moskovitz Remainder Interest Trust,” $1,395,695,354, noncash, “Publicly traded securities,” received 30 June 2025; and “Beneficial AI Foundation,” $150,027. The two sum to the return’s $1,395,845,381 in contributions (ProPublica, Good Ventures FY2024 Form 990-PF; ProPublica, Good Ventures Schedule B).

• Chair of both Coefficient boards. Coefficient’s governance page: “Coefficient Giving LLC is governed by a Board of Managers currently consisting of Cari Tuna (Chair of the Board), Dustin Moskovitz, Elie Hassenfeld, and Alexander Berger. Coefficient Giving Research is governed by a Board of Directors currently consisting of Cari Tuna (Chair of the Board), Dustin Moskovitz, Rakesh Mehta, and Alexander Berger.” The same page says the staff who “investigate and recommend grants or investments” work at Coefficient Giving LLC (Coefficient Giving, Governance). See Open Philanthropy.

• 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. In 2025, we more than doubled that to $351 million, and this year we’re on track to commit over $1 billion.” Coefficient also writes that Good Ventures is its “largest funding partner” and that it is “typically able to recommend giving opportunities to external funders that are 2-5x as cost-effective as Good Ventures’ marginal AI safety funding” (Coefficient Giving, We’re urgently scaling our work on AI and biosecurity; Coefficient Giving, AI safety and security need more funders).

• In her words, on AI (Stanford PACS interview, 6 April 2026). “While there are extremely strong commercial incentives to accelerate AI capabilities, there aren’t comparable incentives to ensure these systems are deployed and integrated safely. I think philanthropy can play a very important role to close that gap.” On what Coefficient funds: “One is increasing visibility into what frontier systems can actually do — that can look like funding evaluations of potentially dangerous capabilities like cyberoffense or biological research, or supporting high-quality journalism about the trajectory of AI development to help inform the public.” On her own change of mind: “I was always glad we had people working on AI safety, but I didn’t fully feel the urgency the way they did,” and “What changed my mind wasn’t a single argument, but rather watching the evidence of AI progress accumulate, and trusting the colleagues who were closest to it.” (Stanford PACS, Philanthropy at an Inflection Point: A Conversation with Cari Tuna, 6 Apr 2026)

• In her words, on method and delegation (same interview). “A big part of how we manage that uncertainty is investing heavily in grantmakers with deep expertise and trusting their judgment because they’re closest to the work and best positioned to spot both the risks and the opportunities.” On the structure: “Instead of hiring our own foundation staff at Good Ventures, we partnered with GiveWell to create a separate project, which eventually became Coefficient Giving.” And: “In 2025, Coefficient directed over $200 million from donors beyond Dustin and me, and more are joining all the time.” (Stanford PACS)

• The founders’ letter, jointly signed. Good Ventures’ letter, signed “Cari & Dustin / Co-founders, Good Ventures,” says: “Cari took the lead at Good Ventures, approaching her learning like a reporting assignment. She studied the history of philanthropy and interviewed hundreds of people across health, science, education and more.” It says the project begun after a conversation with Holden Karnofsky “launched a project, now called Coefficient Giving,” and that the foundation will keep going “until we’ve moved the last dollar out the door” (Good Ventures, Our Story). See Holden Karnofsky.

• Member of GiveWell’s board, beside an Anthropic co-founder. GiveWell’s people page lists “Cari Tuna” under Board of Directors (“Co-Founder and Chair of Good Ventures … She is also the Chair of Coefficient Giving”) on the same list as “Daniela Amodei is President and co-founder of Anthropic” (GiveWell, People). See Anthropic.

• Where the AI line shows up at the evaluators. AVERI’s about page lists “Coefficient Giving” among its funders with Fathom, Halcyon Futures and Good Forever Foundation; the recipients’ amounts and the SFF lines are on the evaluator alliances map and on Dustin Moskovitz’s profile (AVERI, About).

• Political giving, from the federal record. The Federal Election Commission’s itemized individual-contribution search for “TUNA, CARI” returns 60 records dated 2013 to 2019, the occupation field reading “PHILANTHROPIST” (employer “SELF-EMPLOYED” or “SELF”) on most, and “PRESIDENT” with employer “OPEN PHILANTHROPY PROJECT” or “EXECUTIVE DIRECTOR AND PRESIDENT” with employer “GOOD VENTURES” on others, city San Francisco. The largest filed receipts are $2,500,000 to LCV Victory Fund (22 Aug 2016), $2,500,000 to For Our Future (19 Aug 2016), $1,250,000 to MoveOn.org Political Action (26 Aug 2016) and $653,480 to Real Justice PAC (9 Mar 2017) (FEC, individual contributions, TUNA, CARI).

• Social accounts: none verified. Coefficient’s page on her links only the organization’s X account and LinkedIn, and Good Ventures’ pages link no personal account. No handle is added.

Social Persona / Impression Management

Immediate impression: The reporter who became the donor. A former Wall Street Journal reporter who says she approached philanthropy “like a reporting assignment,” who chairs the boards and rarely speaks. The footprint is a governance page, a founders’ letter she co-signs, a 2025 summit appearance and one printed interview.

Energy: Chair, not operator. She delegates grant selection to the adviser’s staff and says so; Good Ventures has no staff of its own. She appears in the record mostly through filings, boards and a handful of talks.

Impression management strategy: The learner under uncertainty. Her interview: “That freedom isn’t comfortable, exactly — you’re constantly making decisions you can’t fully defend yet,” and on her own change of mind, “I didn’t fully feel the urgency the way they did.” First-hand exhibit: the Stanford interview is the one document in which she states her AI view at length (Stanford PACS). Good Ventures’ speaking page lists her 2025 remarks at the Gates Foundation’s 25th anniversary and a Stanford fireside chat, with no transcript (Good Ventures, Speaking and Writing).

Forensic Archetype Comparison

Psychometric Assessment

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

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

Dark Triad (evidence-bound):

MBTI: ENFJ (“The Protagonist”): extraverted feeling in service of a method, convening other donors and mentoring them. First-hand rationale: “Being able to pay that forward has become one of the most personally meaningful parts of my work” (Stanford PACS). Observation of the public role only; a style inference, not a test result.

Threat Assessment

Score rationale. The receipts: Coefficient’s governance and team pages, Good Ventures’ Form 990-PF with Schedule B, the Stanford PACS interview, Good Ventures’ letter and GiveWell’s people page. Attack 7, defense 9, durability 8, hazard 80.0. Peers: Dustin Moskovitz (8/10/9, hazard 88.0), the co-principal: a point above on each, because his name is also a named Funder of the Survival and Flourishing Fund’s 2026 round and an Anthropic investor on the record (attack), his Asana voting power is on an SEC filing (defense), and the foundation’s filed capital is his trust’s (durability). Good Ventures (7/9/8, hazard 80), the foundation she chairs: level on all three, since the person and the organization read from the same return. Holden Karnofsky (8/9/8, hazard 82.0), who allocated the grants: a point above on attack because he chose them where she chairs the board that governs the choosers. Luke Muehlhauser (7/8/7, hazard 72), the AI-governance grantmaker: level on attack, a point below on defense and durability because he is staff and she is the principal on the filing. Attack 7 is the number most open to challenge: the records document a chair, and she says the selection is the staff’s.

Score basis. ATK 7 is leverage: Chair of the Board of Managers of the grantmaking entity and of the research entity, President and Board Chair of the foundation that paid $306,706,819 in grants in the year (Coefficient Giving, Governance; ProPublica, Good Ventures Form 990-PF). DEF 9 is insulation: a private foundation she chairs with no compensation to her, capital documented as a trust’s, no donor to answer to (ProPublica, Good Ventures Schedule B). HP 8 is durability: standing institutions with multidonor pooled funds, offset by a stated spend-down (Stanford PACS).

Alignment Analysis

Stated alignment: Do the most good, under uncertainty, and give the money away in her lifetime. In her words: “The goal is ultimately to do as much good as we can, and that means funding work across the risk spectrum” (Stanford PACS). Observed alignment: Fund, through one adviser and one foundation, much of the apparatus that defines what “safe enough to ship” means, while chairing the boards of the funder and the adviser and serving on the board of the recommender GiveWell beside an Anthropic co-founder (Coefficient Giving, Governance; GiveWell, People). Gap assessment: Small and stated. She says delegation is the method (“trusting their judgment”) and that her own urgency on AI came late and from colleagues. What the record cannot settle is how much of the AI line is Good Ventures money, since the adviser’s pages give totals without a donor split, and who contributed the $150,027 Beneficial AI Foundation line.

Interest: Documented, never motive, never a verdict. What each position would pay the institutions she chairs, with her best case.

1. Position: philanthropy should close the gap between commercial incentives for capability and incentives for safe deployment, including by funding evaluations of dangerous capabilities (her words: “funding evaluations of potentially dangerous capabilities like cyberoffense or biological research,” Stanford PACS). What adoption pays: the grantees on Coefficient’s lists, among them the evaluators the labs use, whose amounts are on Dustin Moskovitz’s page and the evaluator alliances map; Coefficient commits “over $1 billion” on its AI line in 2026. Best case: hers: “there aren’t comparable incentives to ensure these systems are deployed and integrated safely.” Others who gain: external donors, whom Coefficient says it can steer to opportunities “2-5x as cost-effective as Good Ventures’ marginal AI safety funding” (Coefficient Giving).
2. Position: build shared grantmaking infrastructure that other donors can plug into (her words: “Coefficient Giving is built as infrastructure that other donors can plug into”). What adoption pays: the adviser she chairs becomes a multidonor institution, with pooled funds that Coefficient lists; its AI safety pillars are “visibility, safeguards, and capacity” (Coefficient Giving). Best case: “donors are less interested in reinventing the wheel.” Others who gain: every donor who avoids building a foundation, and every grantee that now has one more funder.

Alliance Map

Several of these run at the same time. She chairs the foundation whose adviser funds the evaluators, chairs the adviser’s two boards, and sits on the board of the recommender both descend from, beside the president of a lab the same evaluators grade. The overlap is the finding, not a coordination claim. See the small world.

Convergent Drive Classification

Self-preservation: Survives every transition by holding the chair, not the staff job. First-hand: “Cari Tuna (Chair of the Board)” on two boards (Coefficient Giving, Governance). Goal preservation: Funds the institutions that define the goal, through an intermediary, so the goal is protected by the board before it is debated. First-hand: “we partnered with GiveWell to create a separate project, which eventually became Coefficient Giving” (Stanford PACS). Resource acquisition: Converts a household fortune into a standing grantmaking institution that now advises other donors. First-hand: “In 2025, Coefficient directed over $200 million from donors beyond Dustin and me” (Stanford PACS). Self-improvement: From reporter to donor to chair. First-hand: “approaching her learning like a reporting assignment” (Good Ventures, Our Story).

Subject is not an AI system. The drives appear anyway — in the chair 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.

Public footprint: Coefficient Giving, Cari Tuna · Good Ventures governance · Stanford PACS interview.

Sources: Coefficient Giving, Cari Tuna; Coefficient Giving, Governance; Coefficient Giving, We’re urgently scaling our work on AI and biosecurity; Coefficient Giving, AI safety and security need more funders; Good Ventures, Governance; Good Ventures, Our Story; Good Ventures, Speaking and Writing; ProPublica, Good Ventures FY2024 Form 990-PF; ProPublica, Good Ventures Schedule B; The Giving Pledge; GiveWell, People; Stanford PACS, Philanthropy at an Inflection Point; AVERI, About; FEC, individual contributions.