Since August 2, 2026, every text written by Claude carries an invisible watermark. Within two weeks, the internet was full of predictions about plummeting rankings and Google penalties. But what do we really know now? And how can you best use AI for your content?

Invisible watermarks… eh, what?

Anthropic marks the output of Claude models launched on or after August 2, 2026. Text gets an invisible watermark and files get a digital origin label. The reason for this? That is the legislation of the EU AI Act. This states that providers of AI models are obliged to mark generated output in a machine-readable way. Anthropic, OpenAI, Google, Meta, Microsoft, and Mistral all signed this code of conduct. 

A few caveats: a found watermark only proves that Claude was involved in the process, but not that it conceived the text. A watermark check is something very different from an 'AI detector'. The mark is attached to pieces of phrasing; a language model chooses the options for the follow-up with each word. Slightly rewriting to polish away the watermark makes no difference and is also wasted effort. Moreover, the watermark is actually nothing new. Google has been watermarking images with SynthID since 2023 and text from Gemini since 2024.

Does Google use the watermark as a ranking signal?

At this moment of writing, no source indicates that Google evaluates pages with an AI origin differently. Google looks at the so-called E-E-A-T: Experience, Expertise, Authoritativeness, and Trustworthiness.

• Experience: does the page contain practical / personal experience with the topic? (think of reviews, for example).

• Expertise: are you an expert in the field, with the right qualifications, knowledge, and skills.

• Authoritativeness: are you also recognized by others as a reliable source? (backlinks, reviews)

• Trustworthiness: is the content honest, accurate, transparent, and secure?

So, basically, it doesn't matter whether a freelancer writes 500 generic pages with little added value, or whether these texts are generated by AI. AI only makes this faster and cheaper, but as a result, you will significantly compromise on E-E-A-T.

How do you use AI smartly for content?

At Aizy, we believe in the combination of our own AI software with human expertise. And this exact approach is also the success formula for content with AI. A dream scenario of this collaboration: a strategist determines the topic, search intent, and structure. AI speeds up research, setup, and the first draft. A specialist adds their own data, customer examples, and judgment, checks every factual claim, and removes what contains no information. 

4 practical content & AI tips you can start using today

1. Add your own data to content: customer figures, test results, practical examples. Authentic content that Claude, ChatGPT, or any other platform cannot possibly retrieve from other sources.

2. Replace stock images with your own screenshots. Dashboards, graphs from your own data, and photos of the team are unique visual materials and also carry your origin data. This is something we at Aizy also strongly focus on: even though we are true AI ambassadors, the majority of the photos on our website and in our advertisements are of our teams in the office.

3. Write about the questions your customer service or sales department receives. Those questions are rarely in a keyword tool, while they hit exactly the search intent where conversion lies.

4. Update existing pages instead of publishing new ones. One updated article with fresh figures structurally performs better than a third variant on the same topic.

Would you also like to move away from copy+paste texts from AI platforms and get truly focused on SEO and content? Then contact us. Our specialists will gladly tell you more about our approach.