AI-generated content is becoming easier to create, but regulators are also pushing for greater transparency.

On 10 June 2026, the European Commission published its Code of Practice on marking and labelling AI-generated content. The relevant transparency obligations under Article 50 of the EU AI Act started applying on 2 August 2026.

This has increased attention to AI watermarking, provenance metadata, and other ways to identify AI-generated content.

However, one important distinction matters: the EU requires transparency in certain cases, but that does not mean every piece of AI-generated text must carry a visible watermark.

What Does the EU AI Act Require?

Article 50 focuses on transparency around AI-generated and AI-manipulated content.

The European Commission says providers must design certain AI systems to make AI-generated or manipulated content detectable through machine-readable marking. In specific cases, deployers must also disclose that content was artificially generated or manipulated.

The rules go beyond simple text watermarking.

The exact obligation depends on the type of AI system, content, and whether the organisation is acting as a provider or deployer.

What Is Claude Doing About Watermarking?

This is where businesses need to be careful.

Anthropic’s current official Transparency Hub says Claude provides text-based outputs and that Anthropic continues to explore watermarking technologies with industry and academic researchers.

So, as of August 2026, there is no basis to state as a confirmed Anthropic product feature that every Claude text response carries an invisible watermark.

What Could AI Watermarking Look Like?

If text watermarking is implemented, the basic idea is to embed a statistical pattern into generated text.

The text would still look normal to a reader, but a detection system could potentially identify patterns associated with a particular AI system.

For files such as images, another approach is provenance metadata. This can record information about how a file was created or modified using standards such as C2PA.

However, these technologies have limitations.

Can AI Watermarks Reliably Detect AI Content?

Not completely. A watermark should not be treated as proof of who wrote or edited the final content.

For example:

• Heavy editing can weaken a text watermark.

• Short text may not contain enough information for reliable detection.

• Metadata can be removed through file conversion or screenshots.

• Content created by older or unsupported systems may have no watermark.

• AI can also process or edit human-written content.

This means AI detection and AI watermarking are not the same as proving authorship.

What Happens Next?

The most important date has already passed: 2 August 2026, when the relevant Article 50 transparency obligations began applying.

The European Commission is now moving from creating the rules to implementing and enforcing them.

The Commission's guidance makes one thing clear: no single technology solves AI content transparency in every situation. The Code of Practice provides practical measures for marking and labelling AI-generated content, while organisations remain responsible for meeting the applicable legal requirements.

So the likely direction is not simply “every AI article will have a watermark.”

Instead, businesses should expect a combination of:

• Machine-readable content marking

• Metadata and provenance

• Content labelling

• Detection systems

• Human review where required

• Internal AI content policies

Does AI Watermarking Affect SEO?

Not directly. Google has not officially stated that a watermark on AI-generated content will improve or reduce search rankings.

For SEO, the more important question is whether the content is:

• Helpful

• Accurate

• Original

• Relevant to the search intent

• Supported by reliable information

• Reviewed and improved by people

Therefore, businesses should not create content simply to pass or avoid an AI detector.

The better approach is to use AI where it improves the workflow while keeping research, fact-checking, expertise, and editorial review at the centre of the process.

Why Content Quality Matters More for AI Search

AI search adds another layer to the problem. Your content is no longer competing only for Google rankings. It may also be evaluated as a potential source by ChatGPT, Perplexity, Google AI Overviews, and other AI search systems.

That makes content quality even more important.

A strong AI-search content strategy should focus on:

• Clear answers to specific questions

• Original research and insights

• First-hand expertise

• Strong topical coverage

• Clear headings and structure

• Reliable sources

• Accurate and up-to-date information

• Content that can be easily understood and cited by AI systems

How Vryse Can Help With AI Visibility

This is where a service such as Vryse becomes relevant.

Vryse isn't about hiding AI usage or manipulating watermark detection. It focuses on helping brands improve and measure their visibility across traditional search and AI platforms. Vryse positions its platform around tracking how brands appear across Search and LLMs in one dashboard.

Its content service can support this strategy by creating content designed around:

• Real search intent

• Strong topical coverage

• Original and useful information

• Clear content structure

• Entity and semantic relevance

• Questions buyers actually ask

• Content opportunities identified from AI visibility data

The process can work in a simple cycle:

The process starts with tracking AI visibility, identifying content gaps, creating better content, measuring AI mentions and citations, and continuously improving the strategy.

This is more useful than focusing only on whether a piece of content was created with AI.

Vryse also publishes guidance around AI SEO, content structure, semantic coverage, and AI visibility measurement, reinforcing the connection between content quality and AI search performance.

Final Thoughts

The EU AI Act is moving AI transparency from a voluntary discussion towards a formal regulatory requirement. The key date is 2 August 2026, when the relevant Article 50 obligations began applying.

But businesses should not reduce the issue to AI watermarks alone. The bigger shift is towards transparent AI use, verifiable content, stronger editorial standards, and better provenance.

For brands competing in AI search, the priority should be simple: create content that is useful enough for people to trust and strong enough for AI systems to understand, mention, and cite.

That is where a combination of high-quality content creation and AI visibility tracking can make a difference. Vryse supports both sides by helping brands create content for modern search and measure how it performs across AI platforms.