Programme Thesis
Coefficient Giving's (formerly Open Philanthropy) Career Development and Transition Funding exists to grow the talent pipeline for reducing global catastrophic risks (GCRs), specifically advanced AI risk and catastrophic biological risk. It funds individuals at any career stage who need financial runway to build credentials, domain knowledge, or professional positioning toward a GCR-relevant career — covering everything from grad school fees to self-study stipends to internship income replacement. The underlying theory of change is that talent, not capital, is the primary bottleneck in AI safety and biosecurity, so removing financial barriers to career transitions produces outsized civilisational impact.
Selection Criteria
• PRIMARY GATE — GCR relevance: The proposed activity must credibly advance a career oriented toward reducing risks from advanced AI systems or global catastrophic biological risks (GCBR). Other long-term future causes are eligible but lower priority.
• Career capital specificity: Reviewers want a concrete, bounded activity (degree programme, internship, self-study curriculum, certification period) with a clear timeline and budget — not an open-ended 'keep doing research' request.
• Execution credibility: Does the applicant have the background and track record to complete the proposed activity and convert it into GCR-impact? Prior relevant work, degrees, or demonstrated self-direction all matter.
• Causal chain to GCR reduction: Applicants must explicitly articulate how the funded activity reduces AI or biosecurity catastrophic risk — the link cannot be left implicit. Reviewers apply an importance-neglectedness-tractability lens.
• Counterfactual need: Would the applicant realistically pursue this career path without the grant? Financial need is considered, especially for LMIC applicants or those in low-income career stages.
• Activity type: Grad study (MSc, PhD, law school, MPP), unpaid internships, self-guided curriculum, independent study/research periods, postdocs, professional certifications, and career exploration stints are all explicitly in scope. Long-term open-ended research grants are not the intended vehicle.
• Career stage: Any — from final-year undergrad to mid-career professional to professor on sabbatical. No age or nationality restriction (LMIC applicants are actively welcomed given the counterfactual funding gap).
Past Winners / Cohort Profiles
Open Philanthropy/Coefficient Giving does not publish a public list of individual grantees for this programme. From the programme's own example archetypes (drawn from their EA Forum updates and programme page), funded profiles cluster into five types: (1) STEM researchers transitioning into technical AI safety or ML interpretability — e.g. physics PhDs doing self-guided ML curricula, software engineers adding infosec credentials; (2) biosecurity career-switchers — e.g. computational biology PhDs pivoting to DNA synthesis screening, biologists doing unpaid internships at biosecurity think tanks; (3) policy and law entrants — e.g. individuals attending law school or MPP programmes targeting GCR-relevant government roles; (4) senior practitioners doing exploratory sabbaticals — e.g. management consultants or ML engineers spending 3–6 months mapping how to apply existing skills to GCR reduction; (5) journalists and science communicators entering long-term-future-relevant media. The old Biosecurity Scholarship (now merged into this programme) historically favoured graduate students in quantitative biosecurity fields. There is no public record of named LMIC-based individual grantees, but the fund explicitly notes counterfactual financial need as a positive factor, which benefits applicants from Nigeria and other LMIC countries.
Ideal Candidate Fingerprint
The platonic ideal applicant has a high-ceiling technical background in a field directly adjacent to AI safety or biosecurity (ML, computational biology, genomics, epidemiology, synthetic biology, information security), a specific and time-bounded career transition plan (e.g. 'I am starting a PhD in X in September and need living costs for Y months'), and a crisp one-paragraph argument for how that transition reduces catastrophic risk. They have demonstrated ability to execute independently — publications, side projects, or prior career achievements — and face a genuine financial constraint that the grant resolves. A biosecurity angle benefits from institutional affiliation or a named host (think tank, lab), while an AI safety angle is often self-directed.
Recommended Framing
Eniola's strongest angle is a dual-track biosecurity + AI safety pitch centred on IMPRINT and TOPOLOGIX as safety infrastructure, not just research outputs. For biosecurity: frame IMPRINT (addiction-liability screener) explicitly as a pre-weaponisation detection tool for synthetic CNS drug candidates — the same computational pipeline that predicts reward-circuit encoding probability for therapeutic design also flags whether a novel opioid analogue would produce extreme addiction liability before it escapes into illicit supply chains or is deliberately weaponised, a direct GCBR concern given AI-assisted molecular design. For AI safety: TOPOLOGIX's TDA-driven drug-protein interaction framework provides mechanistic ground truth that current RL-based drug-discovery AI systems lack; without it, generative AI models optimising for drug-target affinity will inadvertently optimise for reward-circuit engagement, producing dangerous candidates at scale. The funded activity to propose is the 12-month pre-MSc independent research and transition period (July 2026–October 2026 plus MSc Year 1 support at MUG Graz from October 2026) — framed as 'converting my CCT computational pipeline into open-source biosecurity screening infrastructure and completing the degree that gives me institutional standing to publish and deploy it.' The endorsements from Berridge, Gershman, Daw, and Mattar — all leading reward-neuroscience figures — should be cited as external validation that the mechanistic grounding is credible, not speculative.
Watch Out
• Non-standard GCR framing: Synthetic-drug-as-bioweapon and AI-drug-discovery-safety are legitimate but non-obvious GCR angles; reviewers steeped in pandemic preparedness and LLM alignment may not immediately see the connection. The application must do the conceptual work explicitly — do not assume the reviewer will make the leap.
• No institutional host for the near-term activity: Open Philanthropy's archetypal funded activities involve joining an institution (grad programme, think tank, lab). Eniola's proposal is essentially continued independent research + MSc preparation, which is fundable but less legible than 'I have an offer from [university] and need living costs.' Naming the MUG/Graz MSc with its October 2026 start date as the institutional anchor is critical.
• Addiction pharmacology is not a canonical GCR cause area: Reviewers will need convincing that reward-circuit work is a biosecurity or AI safety concern, not just important neuropharmacology. The link must be explicit and quantified where possible (e.g. IMPRINT correctly flagged X% of known high-addiction-liability compounds in validation set).
• No biosecurity community track record: Eniola has no publications in biosecurity-specific venues, no think-tank affiliations, and no co-authors from biosecurity institutions. The AI safety secondary angle may be more persuasive because Gershman's arXiv endorsement is a credible signal in that community.
• Grant framing must not look like 'salary for existing independent research': The programme explicitly funds transitions and career-capital building, not open-ended research stipends. The application must present a specific, bounded activity with a clear before/after — 'by the end of this period I will have X credential / Y deployment / Z institutional position' rather than 'I will continue developing the CCT model.'