Could A Cultural And Political Pendulum-Swing Undermine Fashion’s Big Visions For AI?

The Interline doesn’t deal in politics, but as a technology publication it’s hard to avoid the association between electoral issues and AI build-out, which seem to have become synonymous with one another over the last few months – coming to a head in this week’s US midterms, which, in a very real sense, put some strong arguments against AI infrastructure onto local ballots.

Despite a lot of coverage claiming the contrary, there’s little hard evidence to suggest that the pronounced swing in American politics was a direct repudiation of AI, or “big tech”. It’s likely that more immediate and palpable issues, like a rising cost of living, a polarising cabinet, and disruptions to global trade, will have been bigger factors in voters’ decision-making.

All the same, the results have had a hard and fast effect on US markets. The Wall Street Journal reported that the Nasdaq lost $2.5 trillion in market value in a single day (a historic yardstick) as investors positioned for a slowing of the big technology market which, for all intents and purposes, has become the AI market. Which, in a non-trivial sense, means that both voters and institutional and retail investors have either lost confidence in AI, or have seized on an opportunity to enshrine their opposition to it.

Practically speaking, this isn’t exactly surprising. The current US administration has had a very favourable and permissive approach to AI companies, seeing them (correctly, it seems) as the primary engines of US stock market growth and stability, and encouraging those companies to forge quick deals with the communities that would need to accept the construction of data centres to sustain the demand required to sustain that growth.

By contrast, nearly every state where a moratorium on data centre construction was a voting issue saw those restrictions obtain public backing.

This week, then, has provided some concrete evidence of something both close and casual observers have known for a while: that AI has not earned the social permission it needs, in the home market of all three of the frontier AI labs, to justify its outsize influence on stock markets or the large demands it now needs to make for land, resources, and investment to warrant an expected (but shaky-looking) multi-trillion-dollar IPO.

And luxury brands, never ones to miss an opportunity to attune their heritage marketing to stay close to the zeitgeist, have also chosen the last week or so to launch holiday campaigns that are, again, not directly anti-AI, but that place a noteworthy premium on the idea that hand-crafted animation is the antithesis of synthetic, generated content.

Prada, which released a robust behind the scenes documentary about its hybrid hand-animated / live-action ad, seems to have played this riposte successfully. Dior, on the other hand, is under scrutiny for what a group of online detectives seem to think is an attempt to pass off a workflow that might combine traditional animation with undisclosed AI, as an example of historic, pre-digital craft.

For what it’s worth, The Interline has some animation fans on-staff, and there does seem to be something a little odd about the animation that can’t be accounted for by YouTube compression, even though it’s hard to point to anything decisive.

This level of investigation, though, is further evidence that the ambition of both campaigns was well-founded. People don’t like AI. In fact, they dislike it so much that any whiff of it in a luxury campaign looks like it’ll lead to a call for full disclosure and accountability into the production process.

All of which seems pretty clear-cut, until we contrast it with other evidence from this week that, in important ways, consumers seemingly do want to use AI.

Earlier this week, OpenAI consumer product lead Adam Fry told Big Technology (a newsletter The Interline hasn’t come across before, but which seems reputable) that ChatGPT’s volume of “search-like queries” was roughly one-fifth of Google’s, as of this month.

Anecdotally, this mirrors an experience a lot of us have had: “let me ask AI” has become a quick replacement for “let me Google that,” not least because Googling, today, is an exercise in being served an AI overview anyway.

Shopping research published this week also points in the same direction. McKinsey’s survey (which captures data from earlier in the summer) found that 53% of consumers across the markets it surveyed had used AI for at least one shopping task in the preceding three months.

Interestingly, those statistics do not reveal a huge amount of uptake for agentic commerce and checkouts. The data OpenAI published suggest that less than 15% of AI users (i.e. already a subset of the population) have used AI to actually buy things. And this week also saw the CEO of Uber reveal that only single-digit numbers of people had used an AI agent to reserve a rideshare during the company’s busiest period, the Halloween week.

Nevertheless, even if people don’t trust AI enough to make a purchase, actual behaviour appears to contrast with perceived values. Or, to be blunter about it, the public at large seems happy to vote one way and to live another.

In practice, though, the fact that ChatGPT is installed on a lot of phones, and has earned a place as the primary search engine, is likely to pale next to a new legislative mandate to put reins on the AI industry. Which, for fashion’s purposes, means that AI initiatives need to take a new brace of risk factors into account – especially if they are reliant on American-made and hosted closed-weights models rather than open-weights alternatives.

This, a year or so ago, would have been a relatively minor inconvenience. At that time, most AI strategies were focused on the extreme ends of the fashion journey, in early-stage analytical and creative use cases, and then all the way downstream, in marketing and content creation. This pattern is well-documented in our AI Report 2026, and in the accompanying “Arm’s Length AI” essay.

But the final story this week, from Modern Retail, indicates just how embedded AI has become in fashion business models, and how difficult (if not impossible) it would be to extricate it, even if society and incoming senators continue to pursue a campaign of deceleration. That story focuses on Stitch Fix, and their ability to turn the huge volume and variety of consumer and brand data they capture into new ideas, avenues, and assortment strategies for their ten private labels.

This is a prime example of how embedded AI now is in the way fashion companies operate. Stitch Fix itself is a “first-mover in AI,” as we documented in our AI Report 2025, when we interviewed then-CTO Tony Bacos, and this confers the company an advantage in the sense that it has access to data that other brands must assemble piecemeal and from partnerships: insights into not just what gets sold and what gets returned, but how consumers actually interact with, value, wear, and use the products they purchase.

But this strategy is by no means exclusive to companies like Stitch Fix, even if they have an early edge. Even if they do not use AI in customer-facing content (or in animations of seemingly-dubious origin), many companies do explicitly or implicitly use it in multiple go-to-market stages, from planning to supply chain orchestration.

All of which should be a reminder that, for fashion to continue to take big swings at AI, the industry now needs an even keener awareness of who might be caught in that arc, and how strongly they’re prepared to take action to stop it.