HUD、DOGEの政策でのAI利用を認める——その後、証拠を隠蔽
hudは、住宅規制の廃止対象特定にAIを利用したdogeの実態を詳述する100件超の文書を、存在しない「AI特権」を理由に非開示とした。
documents obtained by democracy forward reveal that hud used ai tools, likely including large language models, to analyze and recommend the rescission of housing regulations under the doge initiative. despite public interest and the known risks of algorithmic bias, hud withheld over 100 files related to these activities, claiming protections for 'deliberative ai input' and presidential communications that legal experts say are misapplied or invalid. this opacity highlights a critical gap in us law, which lacks mandates for disclosing ai's role in policy creation, leaving the public unable to audit the 'prompts' or logic driving government deregulation.