Pathfinder / funder / opportunities / Glasswing
Glasswing is described in an EA Forum post as an overlooked governance precedent related to Anthropic’s responsibility to control access to rapidly advancing frontier models when serious harm becomes plausible. Public-record details on its legal form, founders, founding year, and fundraising status were not found in the provided material.
No founder records yet.
This EA Forum post from May 2026 analyzes governance implications of Anthropic's "Glasswing" controlled-access model rollout, arguing labs should not unilaterally decide who receives strategically important AI capabilities. The author demonstrates strong policy reasoning by working through concrete scenarios (democratic backsliding, great-power demands, cybersecurity distribution) and proposes four institutional design criteria including independent regulatory oversight and mandatory AISI access. Composite score of 85 reflects excellent thesis fit (civic infrastructure / AI governance), strong analytical credential (navigates UNGPs, geopolitics, and technical-political boundaries), but zero raise-stage signal since this is commentary, not an organization seeking capital.