What happened: How Claude’s text watermark works

Anthropic is adding invisible watermarks to Claude's text outputs to detect AI involvement, complying with EU AI Act requirements.

Why it matters: How Claude’s text watermark works

Engineers using Claude should understand this when evaluating text authenticity or considering how watermarking affects model performance and output quality.

Watch out: How Claude’s text watermark works

Watermarks work poorly on short text, factual passages, and edited content. Complete rewrites remove watermarks entirely. Detection requires Anthropic's API key.

In the news

How Claude’s text watermark works

Anthropic · Published August 14, 2026 · 3 min read

In 30 seconds

What happened: Anthropic is adding invisible watermarks to Claude's text outputs to detect AI involvement, complying with EU AI Act requirements.
Why it matters: Engineers using Claude should understand this when evaluating text authenticity or considering how watermarking affects model performance and output quality.
Watch out: Watermarks work poorly on short text, factual passages, and edited content. Complete rewrites remove watermarks entirely. Detection requires Anthropic's API key.

Listen to this summary

Read the full article

claude

The patterns behind this

Agentic SRE (Self-Healing Operations)

Each one covers how the technique works, when it earns its cost, and where it breaks.

The Agent Architect

One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.

Weekly email, one-click unsubscribe. We only use your address to send the briefing.