Max Nadeau's Project Tailwind: Recruiting Founders for AI Safety Nonprofits
Coefficient Giving launches initiative to fund neglected AI safety problems through philanthropic startups
1The talent bottleneck in AI safety
Max Nadeau identifies talent as the primary constraint in AI safety, with existing organizations having long lists of unfunded projects and unstaffed engagements with AGI companies and governments. Coefficient Giving's Project Tailwind specifically targets this gap by seeking founders who can transform neglected safety problems into effective organizations. The initiative acknowledges that while funding is available, the critical missing piece is visionary leadership capable of operationalizing theoretical safety research into practical impact.
2Venture-style philanthropy approach
Project Tailwind adopts a 'hits-based giving' model inspired by venture capital, where most grants may have minimal impact but a few transformative successes justify the entire portfolio. This approach deliberately avoids excessive scrutiny of individual grants, focusing instead on fully funding the most promising projects. The $160 million grant to Resolution exemplifies this strategy, backing Geoffrey Irving's ambitious research center focused on aligning superintelligence through AI-accelerated safety work.
3Founder qualifications beyond tech
While technical AI expertise is valuable, Tailwind prioritizes founders with distinctive philanthropic startup skills: discerning theories of impact, willingness to pivot dramatically, and comfort with speculative future risks. Unlike for-profit startups, AI safety nonprofits require leaders who can navigate unmanifested problems and make prescient judgments about emerging risks. This contrasts sharply with commercial pressures that prioritize present-day, revenue-generating safety solutions.
4The nonprofit advantage in safety
Nadeau argues nonprofits can address risks that AI companies systematically neglect due to commercial pressures. Frontier labs face a 'race to the bottom' on safety, prioritizing cheap, quick solutions over fundamental alignment research. Nonprofits operate without these constraints, allowing work on speculative but potentially catastrophic risks like AI-enabled pandemics or loss of human control—areas where corporate willingness to pay remains insufficient.
5Redwood Research pivot case study
The evolution of Redwood Research exemplifies Tailwind's ideal founder mentality. After recognizing mechanistic interpretability's limitations, founders Buck Shlegeris and Ryan Greenblatt dramatically pivoted to AI control research, shrinking to a two-person team for a year. Despite initial criticism, their willingness to endure short-term pain for long-term impact aligns with Tailwind's philosophy that transformative safety work requires such strategic flexibility.
6Unique nonprofit opportunities
Tailwind highlights specific gaps only nonprofits can fill: independent assessment of AI risks across companies, development of costly safety measures commercial entities avoid, and policy engagement requiring neutral analysis. These areas remain beyond the reach of even well-resourced lab teams due to inherent conflicts of interest. The initiative particularly seeks founders who can build organizations to address these structural market failures in AI safety.
7Funding priorities and scales
Tailwind's funding ranges from small grants for unproven ideas to multimillion-dollar commitments for demonstrated successes. While large grants require 'hard-to-fake signals' of founder quality, the program emphasizes scaling organizations based on results rather than pedigree. This tiered approach allows for both high-risk experimentation and substantial backing for initiatives showing measurable impact in neglected areas of AI safety.
8Common founder pitfalls
Nadeau identifies two critical mistakes: insufficient ambition in problem selection and failure to consider implementation pathways. Many founders focus on present-day, tractable issues rather than speculative but potentially catastrophic risks. Others neglect the 'supply chain' of impact—how research translates into real-world changes in behavior or policy. Tailwind seeks founders who can navigate both conceptual and practical challenges simultaneously.
9The outsider advantage
Contrary to conventional wisdom, Nadeau argues that impactful AI safety work often occurs outside major labs. Independent organizations can credibly assess risks across companies, develop costly safety measures, and engage in policy work requiring neutrality—functions inherently constrained within commercial entities. This creates unique opportunities for founders willing to operate outside traditional tech power centers while still influencing their trajectories.
10Tailwind's project categories
The initiative outlines several priority areas for new organizations: technical safety research, policy and governance, ecosystem coordination, and capability assessment. Each category contains specific project stubs like developing auditing frameworks or creating early warning systems for dangerous capabilities. These serve as starting points for founders to develop more detailed proposals, though Tailwind remains open to unlisted ideas demonstrating high potential impact.
11Application strategy for founders
Prospective founders should develop deep visions before applying, whether based on Tailwind's listed ideas or original concepts. The selection process values clear theories of change and demonstrated ability to execute over formal credentials. Successful applicants typically show both technical understanding of AI risks and organizational skills to build teams that can translate research into practical safety interventions across multiple stakeholder groups.
12The urgency of organizational building
With AI progress accelerating, Nadeau frames nonprofit creation as a time-sensitive intervention point. Each new organization expands the field's capacity to address safety challenges that scale with AI capabilities. Tailwind represents a deliberate strategy to multiply impact by identifying and empowering founders who can convert philanthropic funding into structural improvements in how humanity manages existential risks from advanced AI systems.