Fathom

Fathom is an independent, nonpartisan 501(c)(3) organization that aims to advance AI governance and public-interest policies so that AI is “safe, secure, and trusted.” Positioned as a policy-and-implementation oriented nonprofit, it describes its role as providing “actionable frameworks rooted in clear analysis for policymakers, institutions, and the public,” with a stated goal of helping society navigate both the risks and benefits of advanced AI “with seriousness, transparency, and long-term thinking.” Across its materials, Fathom emphasizes moving from analysis to implementation, noting that “while we conduct research, we operate with a strong bias toward action and seek to enact meaningful solutions to bring about an AI future that we actually want to live in.”

Fathom’s strategy is oriented around developing and scaling practical solutions for governing AI in real-world settings, including translating evidence and values into implementable policy approaches. It frames its work in terms of producing solutions and then driving adoption through policy and partnerships—stating, for example, “We translateresearch, shared values, and policy goals into real solutions that ensure AI benefits our society,” alongside commitments to “secure legislative and policy wins” and “collaborate with partners across government, industry, and civil society to build safety and trust architectures.” The organization also foregrounds independence in funding and positioning, stating it is “funded by philanthropists” and does not take donations from corporations (including major technology firms) or from foreign entities associated with “countries of concern.”

A major, explicitly described policy mechanism Fathom advances is an “Independent Oversight Marketplace for AI,” which it presents as a “third-way approach to AI governance.” In this model, companies are incentivized to demonstrate compliance with heightened safety standards via “verification from independent, expert-led bodies,” while state or federal government provides “authorization and oversight” and certification remains “voluntary for companies.” Operationally, Fathom describes a process in which a state authorizes Independent Verification Organizations (IVOs) that are “fully independent from industry,” while the government sets verification outcomes tied to “AI safety or risk levels” using “democratic feedback from the public.” Licensed IVOs then develop technical criteria and verify systems pursuing certification; verified companies receive a “gold standard” signal of a “heightened standard of care,” and the marketplace is intended to advantage safer systems through incentives, with mechanisms for “de-certified” models and “de-licensed” IVOs that fail to maintain independence.

Fathom also highlights convening and synthesis as part of its approach to building institutional capacity for AI governance. For example, its Ashby Workshops convened “policymakers, technologists, enterprise leaders, faith communities, labor advocates, and national security experts” for “facilitated conversation, scenario exercises, and tabletop simulations.” In its 2026 report-out, Fathom argues that “the bottleneck has shifted from technology to institutions,” and explicitly calls to “Build independent verification infrastructure,” emphasizing IVOs as “a concrete framework to create trust without stifling innovation.” The same report frames trust infrastructure as central to outcomes—stating that “Trust is the single most important factor in determining whether AI leads to progress or greater uncertainty—and the infrastructure to build it does not yet exist”—and identifies near-term governance priorities such as child safety and fraud prevention, alongside the idea that “early choices are shaping the long-term trajectory—and delay is itself a choice with consequences.”

Finally, Fathom presents itself as newly scaled with dedicated philanthropic backing. A launch announcement describes the organization as “mapping the ambitions of a broad and diverse society to the specific steps required to move toward its vision for a world with AI,” and states that it is “launching with $25M dedicated to its efforts,” with early work including insights from “polling and ongoing focus groups” on public expectations for government and AI companies. Taken together, the excerpts depict an organization focused on practical, independent, policy-driven AI governance—pairing public-attitude research and multi-stakeholder convening with a concrete governance proposal (independent verification) intended to improve accountability and increase trust in AI systems deployed across society.