Anthropic announced on August 11 that all text generated by Claude now carries an invisible watermark. The watermark survives copy and paste, and appears even when Claude only corrects your spelling. This is to comply with the EU AI Act’s Transparency Code, which took effect on August 2.
If you are using Claude to write emails, web copy, social posts, or anything else a customer or client might read, that content now has a machine-readable signature embedded in it. The marks apply across all Claude surfaces, including the API, Claude Code, Claude Cowork and cloud deployments.
This is not about quality. It is about provenance. Here is what it means if you run a local business.
It does not matter if AI helped. It matters if a person checked.
The watermark does not mean the content is bad. It means a machine wrote it or touched it. That is fine if you edited it, added your own knowledge, and made it yours. That is a problem if you pasted it straight in.
I have seen plenty of AI-written Google Business posts that all sound the same. I have seen AI-generated service descriptions that include things the business does not actually offer. I have seen website copy with American spelling on Australian tradesites.
All of that was easy to spot before. Now it will be trivial.
Google, Meta and others can already see the mark
AI detection software looks for patterns in phrasing, like constructions AI models tend to favour, such as overuse of the word "quietly". The watermark is different. A watermark only helps test whether Claude might have produced or processed the content, and only Anthropic can check for it using their key.
But it is reasonable to assume the platforms where your business shows up, Google, Meta, Microsoft, will have access to those keys or will build their own detection. Even if you use Claude to simply clean up or translate human-drafted content, it could stamp that document with an AI signature.
If you are posting copy that gets flagged as 100 per cent AI-written to your Google Business Profile or Facebook page, do not be surprised if reach drops or if you get less visibility in local search. Platforms reward original, human content. They always have.
The fix is not to stop using AI. It is to use it as a first draft.
I use AI for research, structure, and speed. I do not publish what it gives me. I rewrite it in my own voice. I check it against what I actually know. I add client examples, local details, and anything that makes it specific to the person reading it.
If you are running a cafe in Newstead, your Google post about your new winter menu should not sound like it was written by a chatbot in San Francisco. It should sound like you, mention the actual dishes, and give someone a reason to come in.
That does not take long. It just takes one pass where you read it aloud and ask, would I say this?
AI is a tool. It is a good one. But if you let it speak for your business without your input, you are handing your voice to a machine that has never met your customers.
What to do this week
Prompted by TechCrunch.