Open site menu
PricingDevelopersLabsMethodologyStatusSupportPrivacyTerms
Clean My Copy

Updated August 11, 2026

Which AI tools watermark generated text?

The verified list is shorter than most search results suggest. These statuses use public provider documentation and distinguish text watermarks from image, audio, and video watermarks.

Scan copied text

Confirmed text watermarking

Confirmed by provider

Claude

Supported Claude models launched on or after August 2, 2026

Anthropic says embedded text watermarks apply worldwide across Claude, Claude Platform API, Claude Code, Claude Cowork, Claude Tag, and supported access through AWS, Google Cloud, and Microsoft Foundry. Supported SVG, PNG, and JPG files can also carry signed C2PA provenance metadata. Detection support for users and third parties is forthcoming, but Anthropic has not published a detector endpoint, schema, access program, thresholds, minimum length, pricing, or rate limits.

Read official source

Confirmed by provider

Google Gemini

Text watermarking documented; broad ordinary-text verifier access unconfirmed

Google announced restricted SynthID Detector text support for early testers in May 2025, but its current public verification surface advertises image, video, and audio. Broad ordinary-text verifier access remains unconfirmed. A May 2026 enterprise API announcement does not by itself establish ordinary-text scope. SynthID text watermarking is statistical, not hidden Unicode that character cleanup can remove.

Read official source

Not officially confirmed for ordinary text

ChatGPT and OpenAI text

OpenAI publicly describes text watermarking research, but its current deployed provenance documentation focuses on images and other media.

Microsoft Copilot text

Microsoft documents watermarks for images, video, and audio. We found no official confirmation for ordinary Copilot chat text.

Grok text

xAI documents Grok watermarks for generated images and videos, not ordinary chat text.

Meta AI and Llama text

Meta documents invisible watermarks for generated images, not a deployed text watermark for ordinary Meta AI or Llama responses.

Perplexity, Mistral, Cohere, and other assistants

No official public confirmation of a deployed watermark in ordinary generated text was located as of August 11, 2026.

What this scanner does not claim

A local Unicode scan cannot detect Claude or Gemini statistical watermarks. It can find literal invisible characters and safely clean supported formatting artifacts. A character finding is not proof of who authored the text.