Claude Now Embeds Invisible Watermarks in Text and Files
Left-Hotel904 · reddit · 2026-08-11
Anthropic has documented that Claude now embeds invisible, machine-readable watermarks into all generated content. The watermarking operates at two levels:
- Text: An imperceptible watermark woven into the words without changing meaning or readability.
- Files: Signed provenance metadata on the C2PA open standard for formats like SVG, PNG, and JPG to detect tampering.
Applied at the model level, the signature appears across all interfaces (API, Claude, cloud partners). Models launched after August 2, 2026, will have this from day one, while older models are in a transition period.
More from Models
- First Model to Pass the Video Turing Test: Half of Users Thought It Was Human — Puzzleheaded-King584 · 2026-10-03
- Using System One models in Swift: fast, deterministic decisions via Apple Foundation Models — rxwei · 2026-10-03
- Linux Kernel CVEs Surge From ~500 to 1500+ Per Release, LLMs Blamed for Bulk of the Rise — burny_tech · 2026-10-03
- Developer Complains OpenAI's Coding Model Endlessly Scopes Creeps Instead of Finishing Tasks — DavidWells · 2026-10-03
- Sonnet 5 Spotted in Google Antigravity Backend, Which Still Runs Sonnet 4.6 — brandon_galang · 2026-10-03
- Every's Dev Day chat with Matthew Berman: budget gone the moment he tried Ultrafast — every · 2026-10-03