What we observe
Invisible characters, Unicode anomalies, homoglyphs, technical structures, metadata and provenance information actually present in the content.
After Algorithm
About AI and detection
After Algorithm does not try to guess arbitrarily whether a human or an AI wrote a piece of content.
What we verify
A watermark or provenance is labelled verified only when a documented, compatible mechanism can actually perform that verification.
What we refuse to claim
Clean Unicode does not prove human authorship. An anomaly is not automatically a watermark. C2PA presence is not the same as a valid signature until cryptographic verification is performed.
Our goal
Make technical signals understandable and actionable without turning probability or anomaly into an accusation.
AI markers: what can actually be verified
Not all markers are visible or built in the same way. AfterAlgorithm separates deterministic observations, statistical mechanisms and provenance information so that a clue is never turned into a verdict.
Anthropic says that some Claude models launched on or after 2 August 2026 support text watermarks and, where the format supports it, signed C2PA provenance metadata. Anthropic also says technical detection details will be published later. AfterAlgorithm therefore does not label a Unicode anomaly as a verified “Claude watermark”.
Official Anthropic documentation ↗