The Edit is our weekly show, where Social Editor Grace Robinson quizzes editor-in-chief Ben Hanson on some of the most significant fashion and technology stories from the past seven days.
This week on The Edit, Grace Robinson and Ben Hanson discuss the ASOS cyberattack and the exposure of customer search data, Amazon’s About You profiles and the price of personalisation, nature loss as a fashion sourcing and business continuity risk, OpenAI’s expansion into virtual try-on and business software, and the new Microfibre Action Alliance.
This show is available in video format, embedded above, as well as audio, on Apple Podcasts and Spotify.
For deeper interviews with brand and technology leaders, our prestige interview show releases on Thursdays everywhere you find podcasts – and will be coming to YouTube soon.
Subscribers to our mailing list will receive notifications for both shows.
Grace Robinson: Welcome to The Edit from The Interline, the show where we run a quickfire analysis on our pick of the most important fashion and beauty technology stories from the last seven days. I’m Grace, the Social Editor, and I’m joined by Ben, the Editor-in-Chief. Together, we have around twenty-five minutes to give you our analysis on the stories that we think really matter.
Ben Hanson: Hey, Grace. Good to see you again, as always. Starting to feel like autumn. I don’t know if it’s the same down in London. Up in Manchester, it’s cold, it’s wet, leaves are falling off.
Grace Robinson: Yeah. It feels like we’ve had the first really cold days, so it feels like the seasons have actually changed now.
Ben Hanson: Good. Well, the news headlines haven’t, in a way. There’s still some common themes that we’re going to hear from today. Actually, unique ones. I think we’ve got a pretty diverse set of stories ahead of us. So let’s get rolling.
Why the ASOS breach goes beyond contact details
Grace Robinson: Yes. So the first story that we have to talk about is the ASOS cyberattack. Earlier this week, anyone who had the ASOS app on their phone will have received a message directly from the hackers. Essentially, they said that they’d compromised the Snowflake instance, which, as the BBC outlined, Snowflake is a popular data storage and analysis company. The cybercriminals essentially used a special AI platform to access Snowflake and then manipulate it.
At first, when ASOS spoke on this, they weren’t sure if personal data of users had been released. But now, later communications from ASOS have said that they have. The hackers accessed things like emails and addresses and phone numbers. It is very concerning, and it was definitely a tough week for ASOS. We obviously know that cyberattacks are a lot more common now with AI. So I wanted to get your take on this story, and what should brands be thinking about with all of this?
Ben Hanson: Yeah. So this one is interesting. There’s an AI system involved in it. There’s a platform called Simon AI. I’ve done some reading around and some research here. It does personalisation, segmentation, the kinds of things that you would architect on top of big data warehouse, data lake stuff like Snowflake.
That might be the source of the hack, but it’s not the source of a hack because it’s an AI tool, and neither does this seem to be an AI-enabled hack. This sounds, from all of the reporting that’s been done, like an old-fashioned piece of social engineering where somebody persuades somebody to part with login details to a third-party platform. This is not infrastructure that’s owned by ASOS. This is several steps removed from them.
Then, with that access, they are able to move laterally in some cases and start to get access to other systems, and to do something pretty unprecedented, I think, which is sending a message to every customer that has the app installed. Now, the message is obviously very dramatic. I think it implies a level of access that the hackers didn’t necessarily have here. I think if you’re on the receiving end of that, you would probably rightly assume that this meant that absolutely everything on ASOS’s end had been compromised.
The reality seems to be that it is this segmentation, personalisation layer built on top of data architecture stored in Snowflake. Then it’s paired with and integrated to a system called Braze, I think is the name for it, B-R-A-Z-E, which allows for messaging and follow-ups on particular segments. All of this is built on the personal data that ASOS has collected about customers. Then the way in, the door here, is the AI system that they use to run analysis, segmentation, derive insights and stuff on top of those segments, and then the messaging system that they use to send details out from there.
Now, anyone who’s listened to the show for a while has heard me be a bit fatalistic about cybersecurity attacks in my personal life. I’ve talked about the Manchester Airport Group one. It’s a massive one here in the UK that happened just a few weeks ago in September. I fly out of Manchester Airport all the time, park my car there all the time. Just unbidden, one day you get a message that says, oh, your Wi-Fi login and browsing patterns, your car registration plate, your name, your email address, your date of birth, everything else is contained in a breach and a dataset that’s then been extracted from there.
I don’t blame some people for being like, whatever, again. But the interesting part for this is names, email addresses, phone numbers, stuff like that are basic table stakes for this kind of thing. I think that’s where people get a bit fatalistic, is this idea that, fine, my email address is out there, my phone number is out there. It probably already was anyway. What’s the worst that could happen?
The interesting part of this is the data that was stored in Snowflake and that was then used through Simon AI to do segmentation and personalisation includes search history. It includes the way that people have interacted with the ASOS website, presumably the ASOS app and things as well. It’s not 100% clear at this stage if that search history is necessarily assigned to individual profiles. It’s not 100% clear that your name, if you shop at ASOS, your email address, your phone number, is sat in the same row, for want of a better word, as your search preferences.
But search preferences can reveal a lot about people. The BBC has done some good reporting here that you referenced, but the examples they use are things like searching for petite, which can tell you how tall somebody is. Searching for an occasion outfit at a particular time of year could tell you when someone’s birthday is, if you didn’t have the data, or someone’s wedding or something along those lines.
There’s a lot in there that I think people didn’t necessarily realise was being captured, almost certainly didn’t realise was being retained, and didn’t necessarily realise it was being used to segment messaging and the personalised experience that they get on the website and in the app as well. Not that that should be a shock when you think about it, but getting that kind of qualitative data as part of one of these breaches feels very different to people, and I think can start to enrich some of the quantitative data that gets leaked too.
I haven’t shopped at ASOS in years and years and years. I have done in the past. Do I care that somebody has my email address? No. Not anymore. That’s where that fatalism comes in. Do I care that that profile can start to get enriched with some of this qualitative insight that people capture for personalisation? Yes.
This is what, to me, represents the biggest risk surface for brands. The more of that kind of qualitative data they collect, the more detailed the personal profile they build up on each individual consumer, the bigger the risk of these kinds of breaches becomes. Because it’s not just stuff that people are used to seeing out in the personal domain, it’s stuff that they think is more personal.
So the message coming from the ASOS app, obviously unprecedented, was the reason that it got the headlines. The reason I think it deserves the headlines long term is because it’s a bit of a bellwether for just how much personal data brands collect about us, how long they hold it, what they do with it, and the systems that it lives in and the governance that applies to it.
What price do we pay for personalisation?
Grace Robinson: The next story actually reveals how we are living in this increasingly surveilled world, but it does have a slight comical twist to it. This week, an Amazon user went completely viral online because she found a hidden part of Amazon’s website where you can essentially look at all the assumptions Amazon makes about you based off your purchase history.
She found some pretty standard things, like assumptions that she shops from the women’s department or probably owns a Shark robot vacuum. But she also found a pretty funny assumption that she probably has a flat bum as well, based off what she’s bought. This went so viral, and it prompted a lot of Amazon users to go and find this hidden page and see what Amazon has assumed about them.
Now, I went to look for myself on my own account, and actually there was nothing in terms of assumptions because I honestly don’t use Amazon that much. If I do, I use my sister’s Prime account. I was slightly disappointed not to see anything, but it kind of makes me want to use Amazon a bit more. I wanted to get your take on this story, and I also wanted to know if you’ve checked your assumptions and what Amazon is saying about you.
Ben Hanson: Yeah. I’ve been a heavy Amazon shopper for the last twenty-two years. I went back and checked when my first order from Amazon was, and it was in 2004. Back then, revealing a bit about my nerdy history and stuff, I bought a lot of graphic novels and things. So I went into this thinking, okay, I have shopped on Amazon for two decades plus. No joke, I think I buy stuff from Amazon four or five times a week across this household.
Nothing, disappointingly. Very little profile on me. Awareness of an author that I happened to have listened to an Audible audiobook of. I don’t know if it’s because I use a bunch of script blockers for browsing and always have. But, yeah, I was disappointed not to find that out about me.
This one’s an interesting story because Amazon actually did disclose that this About You, as they call it, page existed back in May. There’s an official press release for it that’s dated in May of this year. I think it’s one of those things that just disappeared in the news cycle. People didn’t really think about it, didn’t really care about it.
Everybody is also familiar with the Amazon very blunt personalisation that they do, where you buy a set of screwdrivers and all of a sudden you just get recommendations for screwdrivers all the time. You’re like, no, actually, I had a specific job to do with these screwdrivers, and hopefully they’re going to last me for the rest of my life. I’m not building up a museum of screwdrivers over here. I just needed something for a specific purpose. So I think people dismissed a bit of that.
I think where this story has come to the headlines now is, yes, it is funny if you have a flat ass. In this case, I think the virality of it comes from that and from that user’s admission as well. She’s like, look, I searched for some shaping leggings or something along those lines, so there’s a reasonable inference in there. That bit of it is kind of funny.
But when you start to think about some of the stuff that we just talked about with the ASOS hack in particular, you can build up quite a detailed profile of somebody based on what they search for and what they buy. Yes, you can get the broad-brush stuff. Do you buy from the women’s department? Do you buy from the men’s department? Do you buy kidswear? I.e. do you have children?
Don’t forget Amazon owns Prime TV. They have access to your watch habits. Audible, they have access to your listening habits. Amazon Music, if you happen to use it. You can build up a very detailed, almost psychometric profile of people based on their online behaviour, particularly if you’re one of the biggest companies out there.
Now, in one sense, people know this, and this is a bargain that they buy into. I shop at Amazon for two reasons. One is convenience. Two is that I’ve maintained a wishlist there for a long time. I’ve used it often enough that the recommendations, which seem to somehow be separate from About You, are reasonably useful to me.
But any personalisation you get from a brand or a retailer as a consumer, any personalisation you want to unlock as a brand or a retailer for your consumers, requires this kind of information to be captured and held. You cannot do it any other way, except for agentic commerce, AI-native commerce, which we’ve talked about a bunch of times, where you’re several turns into a query with ChatGPT or Claude, and that then gets handed off to the retailer, and they get visibility into that behavioural side of things there. Fine.
But if you want to capture this sort of stuff as a brand or a retailer, you have to gather this data. You have to hold it, and this is how you deliver it. That, as we’ve just talked about, becomes a huge risk exposure surface.
I think what will be interesting to me to see now, just to thread the needle from stuff we’ve talked about in the past couple of weeks, is we talked about Meta’s Muse, which is still not available outside the US, the always-on agent, and people using it to go and cancel subscriptions they don’t have anymore, for instance, or to negotiate a better mobile phone plan, or other everyday admin tasks.
There’s a big question now about, do people use those always-on agents to start to do returns for them, things that feel onerous that they might not have done for themselves? The other is, do they start to press agents into service of deleting these kinds of profile information? Because very easily right now, I could just go say to ChatGPT, I feel uncomfortable about the amount of information Amazon has about me. Go and sort it out. It will either go manually delete all of those rows, put in a takeout request, whatever it might be.
Retailers are walking a tightrope, I think, with personalisation at the moment. Every time it comes bubbling up through these kinds of viral stories is a reminder that, as a consumer, you pay for personalisation. As a retailer, you pay a price for personalisation as well, but it’s measured in data governance and risk exposure.
Nature risk is a business continuity problem
Grace Robinson: Next, I want to talk about how nature could actually be becoming a strategic pillar for fashion brands. I saw a really great piece in FashionUnited this week, and it was written by Dr Michael Burgass, who is the co-founder and director of Biodiversify.
In the piece, he describes that big fashion brands could actually lose up to 30% of their profit due to nature loss by 2030. This is, of course, because the fashion industry is so reliant on materials that need nature in order to be created, like cotton, for instance. He describes that the solution to this is for brands to start thinking about nature through more of a strategic lens and combining it with other expertise, like supply chain intelligence and stuff like this, rather than it being an add-on sustainability consideration that isn’t taken too seriously.
Apparently, according to the article, Kering is among the first businesses to set valid science-based targets for nature. I wanted to get your take on this because it’s a new way to look at nature through more of a strategic lens, but I wanted to know what you think brands should be thinking about this.
Ben Hanson: So if you think I’m fatalistic about data breaches, you’re going to find me even more fatalistic about the business contingency planning, business continuity impact of climate change. It’s great that brands are committing to environmental targets, ecosystem targets, in order to protect their own inputs, which is what we’re talking about here. We’re talking about raw fibre harvesting and so on. We’re talking about leather, the whole supply chain there as it becomes an input to their products, as opposed to the environment in a holistic sense.
A good measure of this is outside anyone’s hands, right? There is a certain amount of climate change-related, climate transformation, climate crisis, whatever you want to call it, related transformation to the material supply chain that is locked in. It’s not something that brands can do anything about, except for where they are big luxury conglomerates or so on that can throw around their purchasing power to secure their share of a potentially dwindling supply.
Now, everyone’s been conversant with this idea for a couple of years in some extreme cases. The go-to example of the procurement risk that changing climate creates is something like vicuña wool. I think vicuña are a little bit like alpacas or llamas, but they live at relatively high altitudes, which makes that a very scarce material to begin with. It’s only used in luxury products or so on. Climate change is impacting where their food sources grow and is driving those into even higher altitudes, which makes it even harder to collect the wool, makes it even more pricey. You get a pure, simple impact on commodity pricing there.
But you can extrapolate that to the whole cotton supply chain. All it takes is a single season for a particular cotton-growing region to have insufficient rain or a deluge of flooding or what have you for the commodity price of that underlying fibre to skyrocket. Overall, I think the prediction here is that progressive soil health erosion, biodiversity loss, all of this kind of stuff leads to just a reduction in availability overall. Net, the fashion industry, the textile industry, every other industry that’s competing for these inputs just has less to go after over time, and some of that is locked in.
Now, there’s an alternative, of course, which is innovative fibres, alternatives, recycled, renewed and things. Those have their scaling issues, which we’ve talked about on the big interview show and stuff before, but could end up being a replacement if the industry takes preferred material status seriously enough to scale those up.
The other alternative is more synthetics, and the industry already uses a ton of synthetics. The majority of textile production overall is still fossil fuel-derived, which has a burden that I know you and I are going to talk about at the end of the show here.
I agree with this, essentially. We’ve said for a while, and I’ve said for a while, that climate change and sustainability are business continuity risks, more than they are reputational risks to me. If you rely on a steady inflow of a naturally occurring fibre, and that is the primary input into your product, and that’s the primary force that governs the pricing of your product, then you’re likely to have something much, much bigger than traditional supply shocks in the very near future.
So, yeah, the environment absolutely is a strategic priority for fashion and is a business continuity priority for fashion as well.
Will AI labs absorb fashion technology’s capabilities?
Grace Robinson: The next story is a very big story for fashion tech startups because this week we saw that OpenAI has released a new feature into ChatGPT. Essentially, now if someone is shopping for different fashion products, once they get the results, they’ll have the option to virtually try on the fashion piece. You can do that by either taking a quick picture of yourself or uploading an existing picture of yourself.
This is a big story for fashion tech startups because it slightly makes a lot of fashion virtual try-on apps a bit obsolete. I’m thinking just of Vybe, for instance, which is an AI-powered virtual try-on extension for different fashion brands. This now seems like it’s going to be cannibalised by ChatGPT.
Additionally, we also saw that OpenAI is really trying to expand its business capabilities of AI as well. We’ve seen that they’ve worked with Ironclad, which is a company that makes software for managing legal contracts and other legal work like this. They’ve been working with internal people from Ironclad to train AI to now complete tasks such as setting up contracts or creating approval processes, and it essentially will now follow the company’s rules.
It seems like a massive leap forward in terms of AI capabilities. I know that OpenAI has also put out a call to work with other businesses in this way. So I wanted to get your take on both of these stories and what you think fashion brands and fashion tech startups should be thinking about.
Ben Hanson: Sure. I think this is a risk to the AI virtual try-on companies that exist out there already. I don’t think it’s necessarily a lock that they will be disrupted or routed and that everything is going to move to ChatGPT, for two reasons.
One is this is the kind of bet that requires people to live in ChatGPT. This hinges on the idea that instead of macOS, Windows and desktop applications, or whatever web browser you use, people will increasingly start to work in AI applications as their primary operating environment. Now, I’ve said before on this show and others that that actually does describe me a bit more these days. I actually do work in Codex, ChatGPT Work, whatever you call it, for a lot of the day. It is constantly open there. But do I use it for shopping? No. We’ve talked about this before. I know you do a little bit, but not to any kind of great extent.
The fundamental difference between this and existing AI virtual try-on things is that this is in ChatGPT’s world and OpenAI’s world, whereas a lot of them are more about embedding it into e-commerce storefronts that the brands already have. So I think it’s competitive, but I don’t think it’s necessarily a straight-shot replacement.
However, a lot of virtual try-on companies, a lot of AI generative, they all rely on the same underlying models. I think this is a strong reminder that if your basic capability is provided by an external party, if it’s provided by a frontier AI lab, it’s much easier for that frontier AI lab to recreate your customer experience based on that than it would be for you to go and train a model to do the underlying piece of it.
You and I talked about virtual try-on last week going more into physical stores, which I think, again, is a hedge against this to some extent. Now, I have not had a chance to try this out. I should do. Maybe I’ll task you with doing the same next week as well. Give it a go and see how well it works. I imagine it probably works as well as a lot of the other solutions that exist out there.
Now, the Ironclad one is a bit more interesting because, on the surface, that seems like nothing to do with fashion, right? Legal contracts, it seems like it’s way outside of our sphere. What it is is a demonstration of the power of computer use, as they call it. Specifically for this one, it’s GPT-6 Astra. The test is turning that model loose and saying, okay, go use the systems that Ironclad provides and see how much of this work you can do end to end.
Now, interestingly, Ironclad have their own agents already. They have their own agentic capabilities built into that system. This is a weird one where they’re selling AI capabilities themselves, but they’re also saying, okay, we think the future might look different. The future might be AI agents actually directly steering the software here.
We talked in previous weeks, I think, about computer use when it came to 3D modelling and things like that, and when it comes to 2D artwork. I have tried a bunch of it for both of those things, and it’s pretty good. If that’s something that you would need to do, if you need to put together 2D artwork, if you need to do some 3D modelling, GPT-6 Astra in particular is very capable of driving your computer to do that.
Now, think about web-based applications and things for the rest of fashion. Think about PLM, ERP, supply chain management, all of those things that you sit down and you open in a web browser tab, and you put in data, or you go through and you are doing an end-to-end workflow. You’re like, okay, this is what we need to do next in material development or so on.
One way of looking at AI is that the AI will be embedded into the system over time, that your PLM solution will have more AI and you will chat to it and you will go, go do this for me. This is more evidence of the longer-term bet, I think, that AI labs are placing on, well, let’s just sidestep all of that, and let’s just have it actually use the software for itself, because it seems to be increasingly capable of doing that.
There’s a famous Apple term, among Apple analysts and consumers. There used to be an app called Sherlock, and Apple once just created a first-party version of it, and it ended up obviating the need for the third-party plug-in or application. That’s why it gets referred to as Sherlocking, where you are a big company, a small developer builds something unique on top of what you do, and then, as the big company, you go, now I’m just going to build that for myself. I think there’s more of that to come.
[Editor’s note: The third-party app was called Watson. Sherlock was Apple’s product, whose adoption of similar features gave rise to the term ‘Sherlocking’.]
Can a microfibre alliance solve the problem at source?
Grace Robinson: For the final story, we’re going back to the topic of synthetic materials, which you just mentioned. This week, it’s been announced that brands like Nike and Patagonia and Under Armour and Lululemon, among other big brands as well, have founded this new initiative that’s essentially meant to reduce microfibre pollution in the apparel industry.
As you’ll know, fibres are responsible for releasing a lot of microscopic microplastics into the environment. These are very harmful to the environment, and they’re also very harmful to our health because ultimately they end up in the oceans, and then we consume fish from the ocean. There are so many negative implications of microfibres.
Now, all of these big sportswear brands have come together, and they’ve started this new initiative called the Microfibre Action Alliance. Essentially, it encourages apparel companies to collaborate on creating real solutions to microfibre shedding. All of the businesses are going to contribute slightly different things. Nike is contributing research, I believe, and Under Armour is contributing data. But ultimately, they want to find a long-term solution and a supply chain solution to microfibres.
I personally think that this seems like a super positive initiative, especially because these brands are competitors, in essence. But I wanted to get your take on it based on what you said before and what the industry should make of this.
Ben Hanson: So I can be really quick about this one. The reason that there are so many synthetic fibres in clothing is that they allow you to do things with the performance, durability, or washability and care characteristics of a garment that you historically weren’t able to do with natural materials, right?
Everybody knows the athleisure boom of the last five, six years or whatever it was that’s now dwindling. If you’re old enough like me, you grew up in the era of spandex and aerobic wear and all of that stuff, all of which was unlocked by the fact that you could make cheap, easy-care, stretchy, durable garments from materials like nylon and spandex. All of these are derived from the same long polymer extrusion stuff that gives you plastic, fossil fuel-derived fibres that have those characteristics.
The microplastics shedding happens at every stage of the supply chain for synthetic fibres. You get microplastics shed throughout mill use, through factory use and cutting. You get it shed through a lot of consumer-end stuff, so washing, care, wearing and so on.
There are ways to mitigate some of that, and there’s a lot of effort that goes into capture. Capturing it and filtering it out through wastewater treatment, capturing it at the factory, capturing it in dyeing and places like that through actually filtering effluent and that kind of stuff. All of that is designed to minimise shedding, right? That’s the wording of this: it’s designed to minimise the microplastic output of synthetic fibres.
You can scale that to a point, and I think I agree with you that this is a positive alliance from that point of view. The only real way to curb microplastic output into the marine environment, into the lived environment, into human surroundings and things, is not to use synthetic fibres. There’s a point at which you need to just look at alternative inputs.
Now, you can shed microfibres from natural materials. Cotton sheds microfibres. It just doesn’t shed microfibres made of plastic that don’t biodegrade and that have those potential health impacts and so on. So, yeah, positive step, but to me, a supply chain solution means finding alternative inputs besides fossil fuel-derived synthetics.
