EU AI Act & Anthropic Text Watermarks: What AI Disclosure Means for SEO, Trust, and B2B Reputation
- Bohdan Lytvyn
- August 18, 2026
RELATED SERVICE AREAS:
KEY TERMS:
As some of you know, August 11 2026, Anthropic has announced that they would embed watermarks on the text generated using this AI. So this invited lots of discussion over the social. Let me give you my perspective — my way of connecting the dots around this question.
I'll start by saying that the problem here is about bringing into the light the very fact of using AI to generate / process the content. So if previously some of us could do it secretly — and use it as a secret weapon against competitors and customers, — now, at least in Europe, people would be obliged to disclose. And the greatest question is how the consumers will react. So as a rule, this reaction is expected to be negative. So let's have several ways of connecting the dots. What can be the perspective of it in terms of decision’s timing, and overall implication.
Perspective 1: EU AI Act Master Key
The first perspective, the most evident one, is the EU AI Act legislation of the European Union about artificial intelligence that has come into force in August 2026 would require the companies that are either providers or integrators of AI, anyone who's doing any kind of job processing of content through AI, need to mark the text as being processed by AI. This is to say — a simple badge, AI generated, AI edited, etc.
The exact wording is not clear as yet, so the AI regulators are, as always, have been a little late on the schedule, so they are yet to prepare the formulation. But the overall idea is that the badge has to be in place. And now this regulation needs to be actually enforceable, they are requiring master keys from the AI providers, like Anthropic, in order to actually verify the AI infringement and to have a rightful stance to impose their fines.
And the fines are significant: fifteen millions in cash and three percent of turnover. This is the master key perspective that says:
We need something to be able to enforce our ruling, and therefore the very AI giants will need to give us the master key.
Perspective 2: Overabundance of AI gen content and the Search
The second perspective is that Anthropic, Open AI (ChatGPT), Google and other providers have long been tackling the problem of overabundance of the AI content, for example, in search or as an impediment to machine learning because you will not normally learn on AI generated text. So this is a time, which coincides with AI regulation in Europe, which invites the visible solution.
Let's mark in the first place and (later) segregate the original from generated content to be able to train the machines on the proper content.
This is the second way to connect the dots and obviously to take use of the moment.
Perspective 3: Google and SEO Balance
The third perspective relates to SEO or search traffic in general. Google is a participant or shareholder, if you like, in Anthropic and others. They might get access to this master key(s) and then try to securely segregate the AI generated from original content themselves for the sake of their own purposes.
This may end up affecting the AI generated content in the search, e.g. favorize, for example, human-generated content. So, let's call this perspective a way to redress the balance from the two years back when things started getting worse, and there was a flood of this generated content. So, the third way of connecting the dots is this:
Google may now try to redress the balance and pessimize, for example, AI-generated content in search results.
From these three perspectives, let's then proceed to the implications of this.
SEO Bad News Spiral
First is that what strikes me is that recently SEO has been bad news. So, the traffic from search mostly fell for a major part of the companies. If you see internal docs, the companies are really stressed about year over year traffic decline.
And, SEO as a channel is at its lowest in terms of investment analysis. It's hard to now say, "This time we invest, and then that's our return." Actually, this ratio is getting worse.
What further exacerbates this problem is attribution. So we have fallen traffic. We have a pessimistic investment sentiment, and also we have an attribution. We cannot tell what kind of traffic brings what kind of returns. This problem is getting worse day by day also.
AI overviews and other things on search bring to what's called zero-click web, i.e. the biggest amount of clicks stay in Google and less come to the companies, which affects mostly the publishers, but also every other kind of company.
And finally, AI assistants, the big promises that have been given, but right now AI assistants’ traffic does not live up to these expectations.
This all creates what I call bad news sentiment.
And there comes an apprehension that there's obviously no good alternative at sight. AI assistants are still only a tiny fraction of search, sales and traffic. But, on a psychological part, if, say, an SEO agency comes and says, uh, "Voilà, the regulator imposes requirements on marking down the content, this will entail additional costs to tell which content is which, and then to mark up — this is a bad news because a business only sees costs but not any perspective of revenue.
Great Doubt and Rollback
This evokes what I call the great doubt. Every piece of news like this and every request from your consultant, SEO agency, or internal team evokes the doubt that's inherent in many financial officers or C-levels that the overall AI bubble may burst someday. This is to say we will not pay our investment back, including time and effort. And this is getting even more evident once you understand that you need to mark up all this AI-generated content that's been published on the websites, including the white papers, through leadership pieces, etc.
The next step, which is expected from here, is that we probably need to roll back or dismantle all the stuff we've been creating over the last two years.
This is hundreds of thousands of URLs for many companies, — in my opinion, it might well be about one third of the whole internet over the two years. This is the great doubt that comes down to saying:
What if we must (someday) close the project of AI-generated content for good, which will be a huge waste?
And this obviously translates into the idea of undermining the perspectives of the whole market that's been created off AI services.
So there will be great revolutionary changes, at least in B2B and all reputation sensitive sectors — a reputation under attack — because in the sensitive sectors, even once you put a badge, "AI-generated" or whatever the formulation you use, it will put you in the light that you are probably not that smart as you wanted to look.
Customer Sentiment Risks
Why exactly does this now become a problem? The first thing is, this creates a problem due to customer sentiment. Customers, as they say, have mixed sentiment or, mostly, bad feelings about the AI-generated content.
Ironically, in private life everybody says "We hate AI”. But when we come to work, we are all in favor of it.
Anyway, customer sentiment towards AI is mostly pessimistic. This means once you put a badge over your content, you will probably lose to your competitor that hasn't got this badge, or, what is worse, has a different one, the one that is yet to produce, but not hard to imagine — "Made without AI". That's what puts you on a different side of the new borderline.
This potentially creates huge shocks over the market, especially in the B2B sphere. And this is a risk factor that's obviously beyond control. And what's bad about it is that this factor is compounding as the AI adoption goes on, as the AI poses a risk to the workplaces, etc. This may well turn out to be a punishment to the pioneers, to those who believed in it in the first place. And now they have to get rid of it through using badges or withstand the risk of actual fines.
Tech Giants' Moment of Truth
There is another side to the problem: the technology companies themselves. The largest platforms and AI companies — the gatekeepers of the emerging information ecosystem — have invested huge amounts in AI. They have little incentive to undermine the technology they have invested so heavily in.
At the same time, the growing volume of low-quality and inauthentic AI-generated content creates a problem for the system on which these companies depend. AI systems need access to high-quality, authentic information. Yet publishers are increasingly restricting access to their content, while the open web is simultaneously being flooded with synthetic material. This creates a potentially difficult situation for the future training and development of AI systems.
We may therefore be approaching a moment of truth for the information ecosystem. How much of the information available to AI systems will be genuinely human-produced? How much will be synthetic? And, perhaps most importantly, how will AI companies distinguish between the two?
There is also a potential conflict of incentives here:
Consumers may become increasingly sceptical of AI-generated material. Technology companies may become increasingly concerned about the quality and provenance of the information entering their systems. Yet those same technology companies have enormous commercial incentives to continue developing and deploying AI.
This suggests that the major technology companies are likely to move carefully. They need to protect the credibility of the AI ecosystem without unintentionally undermining the market they are trying to build. And this brings us to the main question: What is actually putting pressure on the market to change?
Regulation as a Global Driver
Ironically enough, one of the strongest forces driving these changes may be regulation and, in particular, regulation coming from the European Union.
The EU's approach to AI transparency is becoming a reference point well beyond Europe. If major AI companies implement watermarking or other disclosure mechanisms, there is little reason to assume that these mechanisms will be limited to European users.
This is important because regulation can turn what might otherwise remain a voluntary industry practice into a requirement. The European regulatory framework therefore has the potential to influence not only European organizations, but also the global practices of the technology companies that serve them.
What Happens Next
So what might happen next — and what should practitioners and managers be thinking about?
First, I expect the European regulators to become more concrete as further guidance clarifies how AI-generated or AI-manipulated content should be identified and disclosed.
Second, European requirements will increasingly have to interact with national implementation and enforcement. Organizations operating in France, Germany and other EU markets will therefore need to consider not only the European-level framework, but also how it translates into their local regulatory environment, which we will likely witness quite soon.
Third, enforcement and sanctions are likely to be in place sometime soon. I would expect regulators to focus initially on demonstrating that the rules are being taken seriously, particularly among larger or more visible organizations and international players.
But the most interesting question may not be regulatory enforcement at all.
It is a consumer reaction.If consumers begin to associate an AI-generated badge with low-quality or generic content, the disclosure itself could become a reputational signal.
We all have already seen the emergence of "SEO content" as a shorthand for content perceived to be generic, manipulative or simply not worth reading. AI-generated content could create another layer of this perception.
That could be particularly significant in B2B markets. In e-commerce, many interactions are relatively transactional. A customer buying a pair of sneakers may care primarily about price, specifications, availability and delivery. The provenance of the product description may be relatively unimportant.
But in reputation-driven businesses — professional services, consulting, financial services, technology, education and other trust-intensive sectors — the perceived authenticity of information can be part of the product itself, which creates a difficult strategic choice.
Organizations may increasingly have to decide whether to:
- disclose their use of AI and accept the potential reputational consequences;
- avoid disclosure and accept the regulatory and compliance risk;
- or redesign their workflows so that AI is used primarily for research, analysis, coding and internal operations rather than directly generating consumer-facing material.
There is no obviously risk-free option. I already hear a potential response:
"Fine. Let's just move everything to ChatGPT."
But it is more likely to be a question of when rather than whether the same pressures reach the rest of the AI providers.
Workflow and Ethics Questions
This ultimately brings us to a much broader set of questions. What is the appropriate workflow for using AI? Should organizations use AI primarily for internal analysis, research, coding and automation while keeping human authorship at the consumer-facing layer?
Or should they accept the potential reputational and regulatory risks of publishing AI-generated material directly?
And why is the public sentiment towards AI so contradictory?
Why do people express such strong negative feelings about AI in their private lives while embracing it enthusiastically in professional settings?
There is also a deeper question concerning trust and identity.
In ordinary commercial and civil relationships, we generally expect to know who or what we are dealing with. Is the counterpart what they claim to be? Is the information presented to us genuinely theirs? Has some part of their identity, expertise or communication been artificially constructed?
AI-generated content introduces a new challenge to this question.
Perhaps part of the attraction of AI has always been the possibility of gaining a private competitive advantage, using a technology that others do not yet fully understand or cannot easily detect.
If disclosure requirements make that advantage visible, some organizations may feel that something they previously considered an internal capability is being taken away from them.
Meanwhile, the largest technology companies will continue to use AI. Governments will continue to regulate it. Consumers will continue to develop their own (negative) attitudes towards it. And businesses will have to face the world developing between all three.
That may be the real source of frustration surrounding AI disclosure: the organizations that adopted the technology early may now find themselves having to reveal how they use it, while the largest players in the ecosystem continue to benefit from it at a completely different scale.
What to do now?
The decision is yours. If your organization uses AI to generate or transform consumer-facing content, you will need to decide how to respond to the emerging disclosure requirements in the European Union, and how much regulatory, reputational and customer-sentiment risk you are prepared to accept.
The question is whether your (resulting) content remains valuable, trustworthy and commercially credible once its origin / provenance becomes visible.
If you are thinking through these questions and want to discuss how AI disclosure could affect your SEO, organic acquisition or B2B reputation, feel free to get in touch.
Organic Growth rarely fails because of lack of effort. Often, it's a result of misalignment between what customers want, what they need and what they are willing to pay for.
If you want to discuss your SEO strategy, let's chat.
About the author
SEO/GEO consultant with 17 years experience in Organic Acquisition and Findability.
Bohdan Lytvyn
"WASTELESS GROWTH" BOOK AUTHOR
Consulting capabilities
SEO Keyword Research
Keyword research to reduce acquisition risk, identify real client motivations, and design organic funnels based on how customers actually search.
SEO Competitor research
I analyze how competitors design their content and SEO systems to satisfy the same client demand, identifying value creation patterns and structural waste.
