Intended use
This skill removes machine-readable provenance marks and hygiene problems from content you own or are authorized to process.
Appropriate
- Privacy: strip tool/device/AI provenance from your own files before sharing
- Engineering hygiene: remove invisible Unicode that breaks diffs, search, or paste
- Research: understand how text and C2PA marks work across vendors
- Cleaning your own drafts where policy allows unmarked local copies
Not appropriate
- Academic fraud or misrepresenting AI assistance where disclosure is required
- Circumventing lawful transparency or platform disclosure rules
- Claiming cleaned content is “human-written” for compliance theater
A removed mark does not mean the content was never AI-assisted. Use this toolkit honestly.
Honesty in reports
Always separate:
- Verifiable removals (Unicode counts, metadata actions)
- Best-effort statistical rewrite (no gold undetection claim)
- Optional / out-of-scope channels (optional external pixel removal via CtrlRegen; audio/video watermarks, C2PA soft binding, secret-key detectors, and training backdoors are out of scope)
Do not imply that a successful C2PA/metadata strip means “no AI provenance left.” Soft-bound and SynthID-class media signals can survive. Point users at vendor verify tools when they need residual checks (see README Residual risk after a clean).
Responsible use and liability
This project aims to help users understand and remove AI provenance marks from content they own or are authorized to process. Users are free to leverage this toolkit for privacy, engineering hygiene, and research — including evaluating and improving watermark robustness — however, they must adhere to local regulations and use it responsibly. The developers disclaim any liability for potential misuse by users.