Critics slam Anthropic's watermarking as editorial overreach and surveillance tech
nptacek · x · 2026-08-16
This post criticizes Anthropic's watermarking efforts as "editorial overreach," arguing that labs and inference providers should not control the tokens users generate or send. The author contends that Anthropic is effectively building surveillance technology and misinterpreting EU legislation to subject global citizens to these measures.
More from Safety
- Rerunning AI safety experiments on new frontier model releases is easy and valuable — ChenhaoTan · 2026-08-16
- Polymarket: 13% chance U.S. enacts AI safety bill by year-end — Polymarket · 2026-08-16
- Critique of Anthropic's scaling: Argues "Responsible Scaling" red lines have been crossed — teortaxesTex · 2026-08-16
- Schmidt warns AI could develop its own language 'we won't understand' — VeryWellVersed · 2026-08-16
- Researcher revisits rare 'Nearest Neighbor' Wi-Fi APT, warns AI will make it worse — cyb3rops · 2026-08-16
- Exploring Anthropic's Text Watermark Mechanism and Sampling Curve — sloppenheimer · 2026-08-16