Sign in
← AI PulseSep 2, 2026

Policy · news · Single-source brief

Anthropic Releases Tool to Check for Claude-Generated Content Credentials

Anthropic has launched a browser-based tool that allows users to check files for content credentials indicating involvement from Claude, supporting various image and audio/video formats up to 100 MB.

By Illumora Editorial

Source · Sep 2, 2026, 4:10 PM · On Illumora · Sep 2, 2026, 4:12 PM

Rewritten from one allowlisted primary — not independent enterprise reporting. Lanes →

Brief drafted by Illumora’s editorial model from the linked primary source. Ops desk reviews flagged pieces. How we write →

Read the source →Claude Security product — Check if content is made with Claude
Save

Anthropic has released a new tool designed to identify if a file was produced or processed by Claude. This browser-based utility allows users to upload files up to 100 MB in various formats, including JPG, PNG, GIF, WEBP, TIFF, HEIC, AVIF, SVG, DNG, JXL, MP4, MOV, AVI, WAV, MP3, M4A, and FLAC.

The tool operates by checking for a cryptographically signed content credential embedded in the file's metadata. This credential adheres to the C2PA industry standard, which is also utilized by camera manufacturers and photo-editing software for provenance tracking.

Key Points

  • The tool checks for content credentials in files up to 100 MB.
  • Supported file formats include JPG, PNG, GIF, WEBP, TIFF, HEIC, AVIF, SVG, DNG, JXL, MP4, MOV, AVI, WAV, MP3, M4A, and FLAC.
  • The content credential is a small, cryptographically signed note in the file’s metadata.
  • This credential uses the open industry standard C2PA.
  • The tool runs in the user's browser, ensuring the file never leaves the device.
  • It identifies only the content credential, not whether Claude was involved in creating the content or any user identifying information.
  • For text, Claude embeds a watermark directly into the writing itself.

Context

According to Anthropic, Claude marks content it produces in two ways: content credentials for files and watermarks for text. The new tool specifically identifies the content credential, which indicates Claude's involvement in producing or processing a file. This credential does not provide information about the content's creation process or the user involved. Anthropic notes that similar tools are available from other model providers such as OpenAI, Google DeepMind, and Gemini.

Why It Matters

This tool provides a mechanism for users to verify the provenance of digital files, specifically whether Claude was involved in their generation or modification. For builders and users, understanding the origin of AI-generated content is crucial for transparency and responsible deployment, particularly as the volume of such content increases.

What To Do

  • Upload a file to the tool to observe how it identifies Claude's content credential.
  • Note the list of supported file formats to understand the scope of the tool's current capabilities.
  • Review Anthropic’s Usage Policy and Privacy Policy linked on the tool's page.
  • Watch for updates regarding the Detection API for text watermarks, currently in private preview for eligible organizations under EU law.

Keep Exploring

/atlas/claude-family