The AI Policy Barrier That's Silencing Marginalized Communities — And Why It's Worse Than You Think

A new paper by Carter Buckner and eight co-authors, accepted at ACM FAccT 2026, examines the steep barriers that grassroots organizations—especially those serving marginalized communities—face when trying to influence AI policy. The researchers present a case study of their own efforts to apply participatory design (PD) principles to U.S. AI policymaking, specifically focusing on developing AI policy for queer people. They describe engagements with several U.S. policy bodies, highlighting structural obstacles such as the lack of extensive networks, lobbying capabilities, and institutional power that make meaningful participation difficult for these groups.

The paper outlines specific challenges encountered during the participatory process, including resource constraints, mistrust of institutions, and the high expertise threshold required to engage with technical policy documents. The authors then provide practical, actionable recommendations for policymakers and organizers, such as providing financial compensation for community input, using plain-language materials, and building long-term relationships rather than one-off consultations. The study contributes to growing evidence that without deliberate effort to lower barriers, AI governance risks being shaped by only the most powerful voices, leaving marginalized communities underrepresented in decisions that directly affect them.