ELI TYRE

Behavioral Archetype

THE STRATEGIST ON BOTH SIDES OF THE MATCH — Subject is Head of Strategy at Palisade Research, an AI-risk research nonprofit, and, in the words of its team page, “supports grantmaking at Jaan Tallinn’s Survival and Flourishing Fund” (Palisade Research, About). The fund’s recommendation tables list Palisade as a recipient in 2024, 2025 and 2026. This profile scores the position, not the man.

The position is the interlock the research map flags: a person on an evaluator’s staff who also appears, on that evaluator’s own page, in the grantmaking process of the fund that pays it. The pages say what the role is and do not say what “supports” covers. He is also the most public-facing member of the Palisade team on the record: project lead and main interviewer of frominside.ai, a set of interviews with current and former staff of frontier labs, and a poster on X under a handle his employer links from its structured team data. The documents show a long career inside the rationalist and AI-safety organizations that preceded Palisade, as his employer tells it, and a present seat at the junction of an evaluator and its funder.

Essence Indicators

• Head of Strategy, Palisade Research. Palisade’s About page lists him as “Head of Strategy” with the biography: “Eli has been professionally focused on risks from advanced AI since 2015, including working for the Center for Applied Rationality as a curriculum developer and instructor, and doing contract work for Lightcone Infrastructure, the Machine Intelligence Research Institute, and the Berkeley Existential Risk Initiative.” (Palisade Research, About).
• Supports grantmaking at the Survival and Flourishing Fund. The same biography ends: “In addition to his work at Palisade, Eli supports grantmaking at Jaan Tallinn’s Survival and Flourishing Fund.” (Palisade Research, About). The Survival and Flourishing Fund describes itself as “a website for organizing the collection and evaluation of applications for donations” (SFF); its SFF-2026 recommendations page lists its Recommenders, among them “Benjamin Weinstein-Raun (Main Track)”, Palisade’s Head of Research, and does not name Tyre (SFF, SFF-2026). The only source for this role found is Palisade’s page.
• Byline on Palisade’s 2026 fundraiser. The post “Help keep AI under human control: 2026 fundraiser,” dated 18 December 2025, carries the byline “Jeffrey Ladish · Benjamin Weinstein-Raun · Eli Tyre · John Steidley” and says: “We have matching grants from the Survival and Flourishing Fund that will double every donation up to $1,133,000.” and “Right now we have about seven months of runway.” (Palisade Research, fundraiser). The post carries no sentence disclosing his role at the fund.
• Project lead, frominside.ai. The site, which describes itself as “a project of Palisade Research,” credits him: “project lead, main interviewer.” Its interviewees are twelve current and former employees of OpenAI, Google DeepMind and Anthropic, among them Jeffrey Ladish; the site states it was “produced by Lightcone Infrastructure” (frominside.ai).
• His own account. Palisade’s structured team data lists his X account as x.com/EpistemicHope (Palisade Research, About). On 29 September 2026 he posted of the interviews: “I’ve been working almost around the clock, for weeks, to get these interviews recorded and get them shipped.” and “If you want to know what people building AI think about the trajectory and the risks, go check it out.” (X, @EpistemicHope).

Social Persona / Impression Management

Immediate impression: An AI-risk professional of the rationalist tradition who writes in public at length. His own description, 28 October 2024: “I’m a vegan aspiring-rationalist, of the Yudkowsky-tradition.” (X, @EpistemicHope).

Energy: High and argumentative on X. He lays out numbered arguments, as on 25 September 2026: “Extrapolating the METR exponential, we should expect one year time horizons around July of 2028.” (X, @EpistemicHope). He states his updating openly: “I’m gaining a lot of sympathy for Eliezer’s sense that approximately all of optimism of mudding through on alignment was a combination of cope” (X, @EpistemicHope), a reference to Eliezer Yudkowsky.

Impression management strategy: Two registers. The institutional one is the team page, which lists a title and a prior-organizations paragraph (Palisade Research, About). The personal one is the X account, which carries opinions on the field, including a critique of his own movement: “The core pitch of EA was “with careful, but simple, first-order reasoning, and relatively small sacrifice, a normal person can do a huge amount of good.” That turns out to be mostly false. Simple first-order reasoning is inadequate to the task.” (X, @EpistemicHope).

Forensic Archetype Comparison

Psychometric Assessment

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

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

Dark Triad (evidence-bound):

MBTI: INTP-leaning, inferred from the documented public record only. The rationale is the work product: long deductive threads, a stated preference for first-principles argument and an explicit habit of re-deriving his own movement’s premises (X, @EpistemicHope).

Threat Assessment

Score move: none. Calibrated against Ben Goldhaber (4/3/3, hazard 33.3), a program officer who places a funder’s money and whose former evaluator employer is among that funder’s grantees, and Kyle Scott (4/4/3, hazard 36.7), an officer at three evaluators, with a point lower on def than Scott because no seat of his is documented as paid or tenured; below Adam Gleave (6/7/6, hazard 55), who runs the evaluator.

Score basis. Atk 4 is leverage: Head of Strategy at Palisade and a stated role supporting grantmaking at the fund that recommended Palisade $1,842,000 for 2026; the points under a higher number are that the page names no vote or title at the fund and that he holds no release gate (Palisade Research, About; SFF, SFF-2026). Def 3 is insulation: his evaluator states about seven months of runway and matching grants from the fund he supports, and no tenure or ownership is documented (Palisade Research, fundraiser). Hp 3 is durability: the frominside.ai interview site and the fundraiser outlive the seat, and no standard or institution is his own (frominside.ai).

Alliance Map

The overlap is the finding. The same biography places Subject in the strategy seat of an evaluator and in the grantmaking support of the fund that recommended that evaluator $1,842,000 for 2026; the fundraiser that evaluator ran under a byline including his, in December 2025, asked donors to give in order to draw the fund’s 1:1 match, and the post carries no sentence about his role at the fund. Palisade’s Head of Research, Benjamin Weinstein-Raun, is separately a 2026 Main Track Recommender on the fund’s page. The fund’s recommendation tables print track totals, not individual Recommenders’ inputs, and state no recusal (SFF, SFF-2026); Palisade’s page does not say what “supports” covers. The documents show disclosed seats and an unspecified role, not coordination.

Alignment Analysis

Stated alignment: In his own words, risk is near and preparation is urgent: “But this argument seems basically sound enough to put enough probability on the possibility of a near term intelligence explosion to be frantically trying to prepare for it.” (25 September 2026) (X, @EpistemicHope). On a plan for the field: “Plan A is the single most thorough and thoughtful plan that I’m aware of for what the world should do as we approach Superintelligence.” (X, @EpistemicHope). Observed alignment: Consistent. He works for an organization that studies loss of control, leads a series of interviews with lab staff and supports the grantmaking of a fund that finances the same field (Palisade Research, About; frominside.ai). Gap assessment: None documented between the stated view and the work. The structural question the record raises and does not settle is the one in the Alliance Map: an evaluator’s strategist who also supports its funder’s grantmaking, with the role’s scope unstated.

Interest: The positions he argues are that a near-term intelligence explosion is plausible enough to prepare for, and that lab staff leaving publicly “might change history.” What adoption would pay, documented: his employer says it has “about seven months of runway” and advertises a match from the fund; a field that treats near-term loss of control as the central problem is the field that funds Palisade’s research and communications work, and the fund’s 2026 round recommended Palisade $1,842,000 (Palisade Research, fundraiser; SFF, SFF-2026). His own best case is the argument itself: the doubling of AI time horizons measured by METR, projected forward five more doublings, in his words “seems basically sound enough” to act on (X, @EpistemicHope). Other factions served by the same premise: the other evaluators that measure the trend, METR among them, which the argument cites as its evidence and which the fund also lists as a recipient (SFF, SFF-2026).

Convergent Drive Classification

Self-preservation: His evaluator publishes its own runway (“about seven months”) and runs a matched fundraiser to extend it (Palisade Research, fundraiser). Goal preservation: Keeps one thesis, near-term loss of control, across CFAR, MIRI, BERI, Lightcone work and Palisade, per the biography (Palisade Research, About). Resource acquisition: The fundraiser he is bylined on asks for donations to draw up to $1,133,000 in matching grants from a fund he supports (Palisade Research, fundraiser). Self-improvement: Writes publicly about compounding his own skill: “I expect to become much more powerful and effective over the next five years than I did over the last five.” (X, @EpistemicHope).

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

Sources: Palisade Research — About; Palisade Research — 2026 fundraiser; frominside.ai; Survival and Flourishing Fund; SFF 2024 recommendations; SFF 2025 recommendations; SFF 2026 recommendations; X — @EpistemicHope; X — @EpistemicHope, 25 Sep 2026; X — @EpistemicHope, 27 Sep 2026; X — @EpistemicHope, 9 Oct 2026; X — @EpistemicHope, 11 Oct 2026; X — @EpistemicHope, 9 Jul 2026; X — @EpistemicHope, 28 Oct 2024; X — @EpistemicHope, 8 Dec 2023.