The End of Takedown Thinking

For more than two decades, online brand protection has evolved in lockstep with the threats it was designed to stop. As counterfeiters moved online, detection tools became more sophisticated. As enforcement improved, infringers adapted. Each technological advance raised the bar on both sides.

‍

What follows is not a prediction. It is an observation from inside the system.

‍

This analysis looks at how brand protection arrived at its current state: a highly optimized, AI-driven ecosystem focused on detection, takedowns, and activity at scale. It explains why those capabilities mattered, how they became table stakes, and why artificial intelligence has dramatically accelerated the cycle.

‍

But it also surfaces a harder truth.

‍

When protection becomes an arms race, progress is measured by speed and volume rather than by control, trust, or long-term brand integrity. The result is an industry that moves faster every year, while brand leaders remain stuck asking the same question: are we actually safer than we were before?

‍

This piece examines how we got here, and why this moment demands a reassessment of what “protection” really means in an AI-driven world.

‍

Having been involved in Brand Protection for the best part of two decades, we have seen many examples of technology being deployed to help protect intellectual property. Even in the days when it amounted to a series of entirely physical processes – raids on markets, customs searches at ports of entry, physical investigation of the supply chain, technology was at the centre of countermeasures. Labelling, packaging, holograms, inks and a succession of other techniques were all used to help rights owners defend their IP.

‍

These each made the task of the infringer more difficult, but none nor all of them amounted to a “silver bullet”. Copying technology was able to overcome many of the packaging/printing hurdles, even authentic packaging waste was re-used and holograms cloned – the point being that serious counterfeiters are committed and do not give up easily, just because a few obstacles get put in their way. 

‍

The Online Challenge

In 2007, I attended an anti-countefeiting conference in Milan, well populated with globally famous fashion houses and the core discussions were all about the sort of physical issues and countermeasures I have discussed above.

‍

When I asked what consideration these businesses were taking of the potential for online fakes, the topic was waved away - “our customers want to buy our products in our beautiful stores, online is not relevant”, was the response. How the world has changed since then!

‍

Digital commerce went through the roof and with the growth of online revenues came an awareness of counterfeit sales on the internet which led to the rapid growth of OBP vendors and five years later, many of those firms were customers of mine at MarkMonitor.

‍

With the rapid growth of counterfeit sales online, the challenge was all about finding and removing infringing listings and websites, so crawler and scanning technology became the key assets underpinning an online brand protection industry that has now grown to be close to a $1Bn global business.

What do we mean by an arms-race ?

There is a saying in sport that “competition is a race without an end”, which also relates to business – just witness how competing vendors in a given market behave, be it server technology, software solutions, chips or even AI agents themselves. 

‍

I have often opined that, if a serious counterfeiting business has a warehouse full of faked brand name leather good, for example, when we remove all their listings and accounts from eBay, Amazon and Ali-Express will they just set fire to them – of course they won’t!! 

‍

They will adjust their listings to avoid easily searched terms, they will set up new accounts and they will shift platforms to other marketplaces such as Segundamano, Allegro and Coupang, or they may even shift strategies to sell via social media such as Facebook or TikTok Shop.

‍

The point is that these are, in the main, serious large-scale players and they respond to challenges in the same way that a legitimate business would – they get smarter, are more agile and employ technology just like the best online businesses would. 

‍

Expert keyword manipulation, advanced cloaking techniques to deliver different content dependent on the IP address or device type of the enquiry, influencer videos with hidden links to invite-only groups that are not easily found by crawling, bot-driven traffic to push listings up search algorithms – these are all examples of how infringers have evolved to avoid detection and compete in a world where brand protection technology itself has improved.

How does AI impact the situation ?

Without doubt AI has been one of the hottest topics in the last year, whether we are talking about financial markets, entertainment, the future of employment or just day-to-day problem solving.

‍

It does, actually, have huge implications for intellectual property rights in several ways that are not directly related to OBP – copyright, music, movies, authorship, and more. Let’s take a look at how it is playing a huge part in many aspects of the threat to brands and how it can also be deployed to protect them.

How is AI used by infringers ?

Unfortunately, many of the lawful activities that AI is suitable for, such as website building and deployment, coding and image creation are core elements of the online infringers value chain – creating listings and sites, copying images and cloning genuine sites are meat and drink to pretty much all of the popular GPT’s.

‍

This means that the potential for “whack-a-mole” is far greater than when criminals had to do all that work themselves. As fast as sites and listings are removed they are rebuilt, maybe on different platforms or with altered tags and keys. 

‍

A couple of years ago, the possibility existed to make a brand very cost ineffective to attack, simply by forcing the criminals to do a lot of work. The impact of that is now dramatically reduced. In addition, AI is brilliant at image manipulation so packaging and product photos, which are often a source of detection for OBP technologies using computer vision, can easily be adjusted to try and fool those systems. Fake reviews and social media commentary can also be added to counterfeit listings to make them appear more credible. 

‍

AI also opens the door for a whole variation of counterfeit sales where it is used to automate the creation of social channels, related influencer content and then deliver bot-led sales via chat platforms which are far more difficult for IP protection teams to track and enforce against. If we consider the impersonation/online fraud aspect of OBP, then AI has an even bigger role to play – the ability to write perfect English emails, with personalisation based on scraped data, spin up SMS or voice-based outreach, supercharges the phisher’s armoury. Add in the ability to clone websites localise language and vary the content per visitor IP, as described earlier, and you have the perfect scenario for a massive increase in online fraud. 

‍

This is why we see so much public facing action from the banks, even here in faraway New Zealand, to warn consumers of the risks of phishing and other online scams.

How can rights owners and their providers use AI to defend themselves ?

I have written about this previously here. Essentially, many segments of the OBP value chain can be assisted or replaced with AI if used correctly :

‍

Detection – scanning the widest variety of platforms using computer vision and natural language processing to find potentially infringing websites and listings.

‍

Analysis and filtering – one of the biggest challenges for OBP providers historically has been that systems throw up a large volume of false positives (e.g., grey market or second user items) that are unenforceable. With image and text recognition, plus machine learning, the most effective OBP technologies are able to identify enforceable items much more quickly and accurately than human analyst teams ever could. 

‍

Enforcement – working with platform abuse teams for listing removal involves a lot of form completion which is a highly repetitive task and ripe for automation, which some vendors have already undertaken.

‍

Innovative applications – just as infringers are using AI to lure consumers into chatbot discussions, vendors such as Marqvision have developed bots that pursue that dialogue with a view to finding the enforceable listing or site behind the transaction. This is a classic “arms-race” example.

‍

Reporting and measuring – unsurprisingly, brands are keen to understand on the status of their OBP projects. Which platforms and locations harbour the most threats, how successfully are they being handled, how quickly and what remains? This has been historically challenging and costly to deliver for vendors, with major customers often requesting highly personalised reporting and KPI measurement that consumed a lot of vendor bandwidth.

‍

Outcome based measurement - Using AI, a small number of providers are now able to calculate meaningful and customised KPI measurements that focus on the impact of the project, rather than just a catalogue of activity, known as “saturation” or “cleanliness” rate. This gives the rights owner a much more meaningful view of the progress of their project than has previously been available.

‍

Conclusions

The conclusion is not that brand protection has failed, it is that the problem it was built to solve has changed. Artificial intelligence has removed friction from infringement at every level: creation, distribution, impersonation, and deception. In response, protection technology has become faster, broader, and more automated. But speed and scale alone no longer equate to control.

‍

When enforcement is perpetual, success becomes temporary. When measurement focuses on activity, outcomes become unclear. When protection is reactive, brands remain exposed at the moments that matter most.

‍

The AI arms race described here is real, and it will continue. But for brand leaders, the more important question is not how advanced the tools are, but whether they meaningfully preserve trust, equity, and market integrity as brands grow. That requires moving beyond a model optimized for takedowns and toward one focused on governance, visibility, and long-term brand integrity.

‍

Understanding the arms race explains the past. Redefining the objective is what will shape the future.