研究

AI、AGI 与 Agent 的权利、义务、责任与治理——从工具、协作者到可能的未来主体。其中包括最低伦理保护问题:在人格问题有定论之前,人机互动应遵循哪些规范。

What AI systems may do with content: read, summarize, transform, retrieve, train, commercialize, redistribute. The AICR ruleset and the AICL-C content-licensing layer make these rights declarable and transactable.

AI 是否可以学习、学到什么深度、用于什么目的、附带什么义务。被读取不等于被学习;“不禁止”不等于“可学习”。AIRS 与 AILP 把学习许可变成分级、机器可读的光谱。

Agent 如何访问网站、API、数据库、知识库与付费内容:身份、权限、请求、支付、授权、使用日志、安全边界与提示注入防御。

What an Agent may DO once it holds tool or API capability, as distinct from what it may read or learn: action taxonomy, effect vectors, reversibility, blast radius, and multi-action composition risk. AARS gives content permission and action permission separate, independently-evaluable axes.

Who an Agent acts for, where its authority came from, whether it can hand that authority to another Agent, and how much of its internal state a verifier may demand. AADP separates principal from actor, bounds delegation to strictly narrow (never silently widen), and treats inspection as a bounded right, not an unlimited one.

机器可发现、可执行的治理规则:llms.txt、/ai/ 清单、/.well-known/ 政策文件、JSON Schema、签名许可令牌与审计日志格式。本站本身就是一个可运行的示例。

AI 与内容的政治经济学:pay-per-crawl、数据红利、主权 AI 基金、创作者补偿池、分层数据市场,以及从公共知识中提取的价值如何回流到生产它的人。