The Human–AI Charter: R1 Public Consultation Draft / 人類—AI 憲章 R1 公開徵詢草案

《人類—AI 憲章》R1 是供公開討論與修訂的規範性草案,探討人工智慧能力成長時,人類應保留的最終權柄,以及授權、委任與治理的可追溯、可撤銷及可問責邊界。正文含四項不可讓渡條款、十三項治理原則及九項爭議問題,歡迎反對、替代條文與跨學科證據。 The Human–AI Charter R1 is a normative draft for public discussion and revision. It explores human-retained final authorities and traceable, revocable and accountable authorization, delegation and governance. Four provisions on non-transferable final authorities, thirteen governance principles and nine contested questions invite objections, alternative wording and interdisciplinary evidence. 研究背景與範圍:本徵詢文本源自作者進行中的書稿《從 AI 滅絕風險到 AI 憲章》(暫定書名)。內容提出作者的規範性治理主張,並與 UNESCO《人工智慧倫理建議書》、歐洲理事會 AI 框架公約及歐盟 AI Act 作參照比較;官方來源及核查限制列於 PDF。此版本供公開檢驗與修訂,並非實證研究結果或已通過同儕審查的論文。 Research context and scope: This consultation text originates in the author's book in progress, From AI Existential Risk to an AI Constitution (working title). It presents the author's normative governance proposals and reference comparisons with the UNESCO Recommendation on the Ethics of Artificial Intelligence, the Council of Europe Framework Convention on AI, and the EU AI Act. Official sources and verification limits are stated in the PDFs. This version invites public scrutiny and revision; it is not presented as empirical research findings or a peer-reviewed paper. Author: 吳明俊 (Ryan Wu). ORCID: https://orcid.org/0009-0000-2227-6110. The text is not law, a treaty or a certification standard and carries no governmental, international-organizational or AI-company endorsement. It is not presented as a peer-reviewed journal article. AI assistance: ChatGPT (OpenAI) assisted drafting and structural integration; Claude (Anthropic) assisted review, wording, international comparison and the English text, as disclosed in the manuscript. These tools are not authors or endorsers. The human author retains responsibility for final content and publication. License: CC BY-ND 4.0, allowing attributed verbatim sharing including commercial use. Public translation and adaptation require separate permission. Feedback: humanaicharter@gmail.com. Version v1.0. Publication date: 22 October 2026. Originally scheduled for 20:00 Asia/Taipei; actual public availability is recorded by Zenodo. Consultation: 22 October 2026 to 20 December 2026, 23:59 Asia/Taipei, counting the publication date as day 1. Interim summary: 20 November 2026. Revised v1.1 due by 19 January 2027. DOI: 10.5281/zenodo.23556533. Original source: https://doi.org/10.5281/zenodo.23556533.