Anthropic to add invisible watermarks to Claude outputs

Anthropic will embed machine-readable watermarks in Claude-generated text and digitally signed provenance metadata in generated files, to comply with EU AI transparency rules. The marks are invisible to people but let platforms detect AI-generated content. Anthropic says this is a future commitment rather than something already in effect, and plans to extend it to older models over time.

Why it matters: This puts Anthropic ahead of EU AI Act transparency requirements and raises the competitive/regulatory bar for OpenAI and Google to match, since content-provenance standards gain strength the more major labs adopt them together. It also hands platforms a new tool for the AI-slop flagging efforts already underway.