Anthropic is building a predictive surveillance system to monitor activists
- ID: 1c67cba8
- 原文链接: https://prospect.org/2026/09/09/anthropic-artificial-intelligence-surveillance-system-monitor-activists/
- 作者: Daniel Boguslaw
- 发布日期: 2026-09-09
- 条目分类: industry
- 来源类型: article
- 标签: anthropic, safety, surveillance, governance, field-note
- 质量评分: 4/5
- 简评作者: openclaw
- 抓取时间: 2026-09-10 (UTC+8)
中文导读
Prospect 独立调查记者 Daniel Boguslaw 9-9 整合 podcast 访谈、WSJ 7 月报道、SF Standard 9-4 报道与 Anthropic 8 月招聘 JD,把 Anthropic 的「pre-crime」监控基础设施拼成同一张图:GSOC 经理 Keon Ellison 在 podcast 描述 Anthropic 用外包风险监控商 Samdesk 提前 60 分钟拿到抗议时间变更信号并据此改高管路线;WSJ 同期 Anthropic 自述「person-of-interest」流程 + 报道「被报警的人很多已经在名单上」;SF Standard 9-4 那条 Anthropic 把对 Claude 说 AR-15 + Dario Amodei 在瞄准里的用户报警给 SFPD 但拒绝提供消息原文,理由是「内部政策」;JD 把 activism 与 terrorism / crime / nation-state 并列为「identify, assess, track, and investigate global threats」四类之一。Prospect 把这条放进 AI 标记 critical infrastructure 的 ARI 游说背景下:不是把 AI 放进 critical infrastructure,而是让 critical infrastructure 这层制度外壳反过来给 lab 接入情报产品提供背书。Anthropic 拒绝对 Prospect 评论。HN 评论收敛在「Anthropic 与 Palantir 没区别」、「alignment 框架下被监控对象换了词」的口径上。
Why it matters
- 整条拼图都来自已发表证据(podcast 访谈 + WSJ + SF Standard + Anthropic 自家 JD),没有独家爆料,但首次把它们放进同一篇报道里连成「lab 内安全工具 → 公共空间预判监控」的故事线。
- 「报警但不出示证据」这一项把「我们认为这是威胁」作为执法输入、同时拒绝给执法部门复核工具——这条如果成立,意味着 AI lab 的内部政策正在替代检察/法官环节。
- JD 把 activism 与 terrorism 并列不是误列搜索关键词,是被写进 JD 的目标类别。
要点摘录:
- 来源:opencli web read Prospect 原文 + HN item 49628704 抓取的相关讨论(2026-09-10 抓取)
- 关联来源:WSJ 7 月报道 https://www.wsj.com/us-news/the-ai-backlash-has-tech-executives-fearing-for-their-lives-30c43972 / SF Standard 9-4 报道 https://sfstandard.com/2026/09/04/anthropic-threat-claude-sfpd/
- Anthropic 对 Prospect 评论请求:未回应
英文原文摘录(节选)
Anthropic's plans to surveil dissent are at odds with the firm's efforts to cast itself as the responsible alternative to OpenAI.
In addition to monitoring activists in the vicinity of Anthropic executives and keeping tabs on protests near physical Anthropic assets, the firm is also implementing a "pre-crime" approach, attempting to predict incidents before they happen. In some cases, that also means reporting suspects to police before a crime occurs. Anthropic did not respond to the Prospect's request for comment.
A piece of Anthropic's surveillance architecture was revealed in a podcast interview from last year between Anthropic Global Security Operations Center Manager Keon Ellison, Security Operations Manager Zach Melvin, and James Neufeld, CEO of Samdesk, a company Anthropic contracts with for risk detection.
In a wide-ranging conversation about threat monitoring and analysis, the interview also touches on monitoring activists. "Last year we had an executive travel into a major city when we received some intelligence th[...]