The following post is based on information provided by: https://support.claude.com/en/articles/16266773-how-claude-marks-ai-generated-content
Following Anthropic’s alignment with the EU AI Act's Code of Practice on Transparency of AI-Generated Content, all models from Anthropic worldwide will begin to contain machine-readable watermarking in the text they generate.
This watermarking occurs regardless of whether you are using Claude Platform (API), Claude, Claude Code, Claude Cowork, Claude Tag, and really anything Claude. In the not-so-short future, Anthropic will also release methods for detecting this watermarking for users and third parties (other companies).
Now that you have been given a quick brief on this information, I would like to say off the bat that Explofi utilizes, to some extent, the Claude API to create SEO-optimized text for users and has done so without problems until today. As Anthropic is deciding to watermark text content, Explofi's response to this machine-embedded watermark is to effectively remove it from all text that contains it (which would be all text created by Claude).
The process by which Explofi removes this watermark does not utilize Claude or any other production LLMs that are sure to follow Anthropic's footsteps of text watermarking.
Additionally, removing the text watermark will NOT hurt text quality or its SEO benefits; in fact, some tests output higher-quality content.
This is seen as a move that is in the best interests of our users.
Reasons Why Explofi Will Be Removing This Watermark
Explofi's move to remove text watermarks from Claude-generated content is backed by three reasons:
(1) Text generated by Claude does not belong to Anthropic; that text was instead taken from human authors and used to train their models.
(2) The watermarking of AI-generated text could very well have a negative effect on the results people see from AI-assisted and optimized content in SEO.
(3) Explofi does not want its users to be subjected to Anthropic's text watermarking.
