Atlas Computing is a 501(c)(3) nonprofit that maps neglected AI safety risks, sources expert founders, and prototypes solutions to scale human control over advanced AI capabilities.
Atlas Computing is a 501(c)(3) nonprofit that maps neglected AI safety risks, sources expert founders, and prototypes solutions to scale human control over advanced AI capabilities.
People
Updated 05/18/26Board Member
Founder and CEO
Board Member
Design and Product
Software Lead
Chief of Staff
Funding Details
Updated 05/18/26- Annual Budget
- -
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- $1,000,000
Org Details
Updated 05/18/26Atlas Computing is a tech 501(c)(3) nonprofit founded in October 2023 by Evan Miyazono, who previously led the research and metascience team at Protocol Labs. The organization's core mission is to map the neglected problems that societies will need to solve as AI grows more capable, and to ensure those problems get dedicated, expert attention. Atlas operates with a distinctive four-step approach: identify overlooked catastrophic risks from AI, map the stakeholders and funders interested in each risk, sketch the shape of a solution and the skills required, then recruit experienced leaders to build focused organizations around those problems. Atlas sees itself as a bridge organization — it takes responsibility for problems until it can hand them off to someone better positioned to work on them full time. Key initiatives include Hardware-Enabled Governors (HEGs), specialized hardware components integrated into GPUs that enable transparent, privacy-preserving compliance monitoring of AI training and deployment. Atlas coordinated technical de-risking of HEG components in partnership with AISTOF support, with the goal of enabling policymakers and frontier labs to implement accountable compute governance. A second major workstream involves formal verification tooling — building infrastructure to help engineers generate formal specifications and validate that software behaves as intended. This work includes the CSlib project (an intermediate representation for computer science concepts in Lean) and a Specification IDE for users unfamiliar with formal methods. CEO Evan Miyazono also serves as a Research Fellow at Convergent Research, where he is designing two Focused Research Organizations (FROs) to carry forward Atlas's work on formal specification tools and hardware compute governance. Atlas also convenes the field through events such as the FMxAI workshop series and the GSAI Summit, gathering researchers from AI and formal methods communities. The organization raised nearly $1 million at founding, primarily from the Survival and Flourishing Fund and Protocol Labs. The core team includes Evan Miyazono (CEO), Tzu Kit Chan (Chief of Staff, former MATS operations lead), and Alexandre Rademaker (FM Research Lead, with 20+ years in formal methods and prior research roles at Microsoft Research, SRI International, and IBM).
Theory of Change
Updated 05/18/26Atlas Computing believes that some of the most important problems on the path to safe advanced AI are neglected not because they are unimportant, but because no one with the right combination of skills and focus has taken ownership of them. By systematically mapping these gaps, identifying funders and stakeholders, and actively recruiting expert founders to lead dedicated efforts, Atlas aims to accelerate the creation of the organizations and tools that will be needed. Their technical bets center on formal verification (enabling humans to precisely specify and verify AI system behavior) and hardware-enabled governance (building compute monitoring infrastructure that lets policymakers and labs enforce safety norms). The causal chain runs from gap identification, through founder sourcing and prototype development, to the launch of focused research organizations that can work on these problems at scale.
Grants Received
Updated 05/18/26Projects
Updated 05/18/26
Design of an AI-assisted toolchain—comprising tools named Formalize, Construct, and Translate—to integrate into critical engineering workflows for code specification, synthesis, and verification.
Atlas Computing initiative announced under the title “Flexible Hardware-Enabled Governors” on the Atlas Blog.
Discussion
Key risk: As a very small team newly pivoting to org-creation, Atlas may fail to secure the founder pipeline, vendor/regulator buy-in, and capital needed to launch and scale HEG hardware or formal-spec tooling, yielding diffuse convenings and prototypes with limited counterfactual impact.
Case for funding: Atlas targets neglected bottlenecks by mapping “missing orgs” and catalyzing spin‑outs in hardware-enabled compute governance and formal verification, with Evan Miyazono’s Convergent Research FRO pipeline and initial traction on HEGs and CSlib/Specification IDE positioning them to turn policy desiderata into adoptable infrastructure.