Key Takeaways:
- This week, Browser Use – which helps AI agents navigate websites more effectively – raised $17 million in funding. eCommerce stores will begin designing with this in mind, but solutions like this could also lessen the burden of the work for brands.
- The debate over AI’s use of publicly available data for training is intensifying, particularly in the context of the UK government’s proposed copyright exception; allowing generative AI companies to use publicly available content without explicit permission. But industry group Walpole argues that this could impact rights-holders and affect business confidence, particularly in industries like fashion and luxury.
- As a counterforce to AI’s sometimes invasive reach, Cloudflare has released its new AI Labyrinth tool – designed to block web-crawling bots that scrape websites for training data. This defence echoes early tactics like “data poisoning,” aimed at disrupting AI model training by safeguarding copyrighted content, particularly in industries like art and fashion.
- Just how much freedom AI will have in the years to come – from the perspective of legislators of different jurisdictions, creators rightfully wanting to protect their work, and from brands wanting to make the most of AI agents in eCommerce – is an open question.
Companies of every shape and size, cross-industry, are going all out on AI agents. According to Salesforce research, a staggering 93% of CIOs plan to deploy them within the next two years. The Interline has been documenting a glimpse of this evolution over the past few weeks, and fresh evidence suggests the pace is only picking up.
This week, technology company Browser Use raised $17 million in funding thanks to its solution that makes websites easier for agents to navigate. Browser Use’s technology addresses a hurdle in web automation: the frequent changes in website layouts and structures. Traditional AI navigation methods, often relying on image recognition, can be disrupted by such changes. Browser Use’s text-based system was designed to offer reliability and adaptability, ensuring consistent performance even as websites change.
Inevitably, the design of eCommerce stores will begin with this in mind more and more, but solutions like this (something that Browser Use calls a “fundamental layer” to any website) could also lessen the burden of the work for brands – giving them space to focus on growth without the constant worry of having to update their sites to keep up with changing demands.
Browser Use is one of the agents powering Manus, the AI agent platform from Chinese startup Butterfly Effect. Fittingly – or perhaps inevitably, given the name – Browser Use went viral alongside Manus, amplifying its reach in a lateral ripple effect. The Interline conducted some research, prompting Manus to curate a selection of clothing from various brands for a father and his 9-month-old, assembling a shopping cart with stylish yet practical options for a June holiday in Provence, France – allowing for final adjustments before purchase.
The result? A simply-styled but detailed website that included the factoring of Provence’s June weather conditions, and a breakdown of clothing and accessories for each member of the family, budget options, French brands to consider, as well as a shopping strategy.
While The Interline does see the utility this kind of site creation may have for customers, we were expecting more. A core component, in our view, would be to have shoppable links or another path to purchase for the products – which was not provided.
Despite being somewhat limited at present, this type of technology is evolving rapidly and holds exciting potential for consumers seeking convenience and personalisation, without the need to search the web, or brand’s website, themselves.
Browser Use has also likely caught the attention of OpenAI, as it’s a direct competitor to their Operator system, which works in its own browser and can look at a webpage and interact with it by typing, clicking, and scrolling. Operator (though still in research preview) is part of a wider trend where agentic AI innovations are quickly transitioning from the research phase to fully realised products.
As The Interline has explored previously: what happens when bots actually start doing the shopping, reading, and, ultimately, deciding?
But first, these agents need content to train on, and it’s the source of that content that’s causing major concern around the world.
This week, Walpole – an industry group that represents 250 of the top luxury companies in the fashion, beauty, automotive, design and hospitality sectors – spoke out against the UK Labour government’s AI copyright exception proposal. As part of a broader effort to accelerate generative AI development in the UK, the government has introduced a set of proposals tackling copyright concerns, with a key element being a proposed exception to copyright law. This would permit generative AI companies to train on publicly available internet content without needing explicit permission from creators. In return, these companies would be expected to provide greater transparency about the data they use and its sources, while creators would have the option to “reserve their rights” (i.e. to opt out.)
However, head of Walpole, Helen Brocklebank points out that a feasible opt-out system does not [currently] exist, so putting this proposal in place would only burden businesses and rights-holders with extra costs, and that while the cost of compliance with these plans would be considerable, there would also be an impact on wider business confidence.
This proposal is not occurring in isolation, but forms part of a wider, deeply entangled legal, ethical, and cultural battle over the future of generative AI.
So far, most of the outcry around these issues has come from music, publishing, photography and art, but this marks one of the first instances where fashion and luxury are speaking out.
Fashion has largely avoided direct legal battles around copyright and generative AI, even as image generation tools begin replicating iconic visual cues like the Nike swoosh. With copyright protections potentially weakening, it remains to be seen whether the industry will have grounds for future legal action as other copyright cases progress.
At this moment where there is growing enthusiasm about agentic AI and the potential benefits of AI copyright – with many advocating for the rights and protections of AI-generated content – there is also a counterforce coming to the fore.
While there’s growing momentum in favour of AI-friendly copyright reform, there’s also investment with tools that challenge it. Cloudflare’s AI Labyrinth, designed to block web scrapers, adds another layer to the debate over AI data access. It recalls an early strategy known as “data poisoning” developed by researchers at the University of Chicago and profiled in MIT Technology Review, which aimed to sabotage generative image models like Midjourney, DALL E 3m and Stable Diffusion by corrupting data.
Less combative – but equally significant – are brand strategies leaning into direct human connection. Listrack’s 2025 Beauty & Fashion Benchmark Report notes that browsing, SMS, loyalty programmes and other personalised outreach methods are increasingly differentiators for brands.
The Interline recently explored this through a firsthand experience with Ffern, a seasonal fragrance brand. Ffern only offers its scents to customers on its ledger, which – at the time of writing – was full. Joining meant waiting for a message to confirm a spot. When that message arrived, the sense of personal recognition was strong, and for us, a key motivator to purchase.
Even as agentic AI grows more sophisticated, and the balance between ownership, automation, and authenticity grows more complex, brands may want to ensure that human engagement remains part of their roadmaps. Because in our opinion, trust, emotion, and loyalty are still anchored in real relationships.
This tension – between freedom, regulation, and innovation – will continue to play out across jurisdictions, industries, and consumer experiences. As the AI-centric web is being built in real time, whether it empowers shoppers, benefits brands or undermines them, or simply feeds the next algorithm, depends on the choices being made now. And there’s a strong likelihood that, with hindsight, we’ll clearly see how the balance between innovation and responsibility was defined at this critical moment.