What AI Synthetic Watermarking means for your business

Digital Marketing

By Sarita Gold

•

07/09/2026

What AI Synthetic Watermarking means for your business

AI-generated content is entering a new phase. Every single day. The rest of us are just trying to keep up.

AI-generated content is entering a new phase. Every single day. The rest of us are just trying to keep up.

Since 2023, when LLM’s or AI became the new buzzwords, much of the conversation around AI and marketing focused on generation: how quickly brands can create copy, images, video, ideas and campaigns. There has been a huge shift recently, with brands and consumers starting to recognise the “slop” being executed across channels. Neither the copy nor the execution is necessarily on-message or appropriate, and often ineffective. Take, for instance, misleading food brand ads. Every single kebab shop, burger joint or goodness bowl spot has churned out the same standardised looks on its posters. Just search for it and you’ll understand.

Looking to understand how AD Media works with AI?

In August 2026, new transparency requirements under the European Union’s AI Act came into effect, requiring providers of generative AI systems to make generated or manipulated content detectable in machine-readable ways where technically feasible. Major AI companies, including OpenAI, Anthropic, Google, Meta, Microsoft, and Mistral, have signed the associated Code of Practice.

Platinum Asset Management

Anthropic likens it to “a game of Monopoly where the dice are replaced by digits of pi. The moves still appear random, but someone who knows the pattern can work out how they were generated.”

Sasha de Marigny
Chief Brand and Communications Officer/Anthropic

What is Synthetic Watermarking?

Forget the idea of a visible stamp or logo. AI watermarking is closer to an invisible fingerprint embedded in generated content. There are broadly two ways this can work. A watermark can be embedded in the content itself, while provenance metadata functions acts more like a digital passport attached to the file, recording information about where it came from, which tool created it, and, in some cases, how it has been edited. One is baked into the content. the other travels alongside it. The important part for companies isn’t knowing the technical difference between emerging standards. It’s understanding what they represent.

How are AI tools implementing watermarks?

As Anthropic explains, future Claude models will include a text watermark designed to indicate the likelihood that Claude contributed to a passage. The words will still read normally to us, but across a longer passage, Claude can leave a statistical pattern that makes its involvement detectable to someone with the right system. OpenAI is using a different approach for other forms of content. Supported generated images currently include C2PA Content Credentials and SynthID, while supported generated audio can also contain SynthID. The technologies differ, but the direction is similar: it is becoming increasingly possible to identify when generative AI has contributed to digital content. This does not mean that an AI watermark automatically signals poor-quality content to Google. Nor can it reveal whether someone used AI for a small part of the creative process or relied on it to produce an entire campaign.

Anthropic notes that its watermark can indicate likely Claude involvement, but cannot determine the precise balance between human and AI authorship.

Platinum Asset Management

Anthropic’s new approach does not insert strange characters, hidden words or additional tokens into Claude responses. Instead, Claude makes tiny statistical choices between different equally suitable words as it generates a response. Across a sufficiently long passage, those choices form a pattern that can later be tested using Anthropic’s watermarking key.

Different technology. Similar direction.

AI-generated material is becoming increasingly traceable. And for businesses producing large amounts of web, social and marketing content with AI, that makes the human contribution increasingly important.

The bigger problem isn’t AI. It’s AI slop.

We’re all getting better at spotting it. The generic article. The strangely familiar social caption, the website packed with words and tell-all dashes. The tell-tale signs of unconsidered AI content are becoming easier for people to recognise, while the technology used to trace synthetic content is becoming more sophisticated too.

Does that mean Google will penalise AI content?

Google’s current protocols focus on whether content is genuinely useful, original and created for people. The problem is mass-producing content that adds little value, regardless of whether it was written by a person or generated by an AI model. That distinction matters. Watermarking doesn’t need to become an SEO penalty for generic AI content to become a liability. Search itself is already moving towards useful, distinctive and non-commodity information.

What does that mean for your SEO and GEO?

If you run a business, the question shouldn’t be ‘Can AI write this for us?’ It should be ‘What can we add that AI couldn’t produce without us?’

Your experience. Your customers. Your results. Your people. Your point of view. Your research, case studies and understanding of the market you actually work in. That is what gives both traditional search engines and generative platforms something worth finding, trusting and potentially citing.

After all, ‘Good GEO increasingly looks like good SEO combined with content worth citing’. (Danny Sullivan, Google’s Search Liaison – August 28, 2025).

AI should accelerate expertise, not replace it

This is where working with people who understand technology matters. At AD Media, we utilise AI to support research, originate, and analyse to produce different content depending on our clients’ requirements.

Combining human insights, experience, strategy, and creative judgement is essential. Good GEO increasingly looks like good SEO combined with content worth citing.

Our clients don’t need more content simply because AI can produce it. They need content that unmistakably identifies their brand.

AI can help brands move faster. Our job is making sure the work is still moving in the right direction.

Related Articles