About
An independent reference for AI content provenance
AI Watermark Center documents watermarks, hidden characters, metadata, C2PA, and related provenance systems with an evidence standard: we only claim what we can verify.
What this project is
AI Watermark Center is building a public toolkit and research archive for a practical problem: AI-generated text, images, and other media can carry inspectable artefacts, signed credentials, or provider-side watermarks — and those things are constantly confused with one another in search results and product marketing.
The problem we address is not “how to make AI content undetectable.” It is how to inspect what is actually present, cite what providers have documented, and stay silent where independent verification is not yet possible.
Independence
AI Watermark Center is an independent project. It is not affiliated with Anthropic, OpenAI, Google, Meta, Microsoft, Stability AI, Midjourney, or any other model or platform provider unless a later page states an explicit relationship.
Provider names appear here because they are the subjects of public documentation and user questions, including queries such as “Claude watermark remover.” Mentioning a name is not an endorsement, a partnership, or a claim that we operate a vendor’s private detector.
Evidence-based research
Factual pages about providers will prefer primary sources, record access dates, and use the methodology hierarchy: officially documented, independently verified, observed, or unverified. Tools will report what they can re-check on the page.
What Phase 1 is not
This version of the site is the public foundation: architecture, editorial rules, and explainers. It does not yet run watermark detection, hidden-character cleaning, metadata parsing, C2PA validation, or advertising.