TikTok’s New US Algorithm | BoF

Subscribe to Tech Mode with Marc Bain, a deep dive into the most intriguing developments in artificial intelligence and its impact on the fashion industry.

Dear BoF Community,

Welcome back to Tech Mode, your monthly guide to how AI and other technologies are reshaping the fashion industry.

First, PVH Corp, owner of Calvin Klein and Tommy Hilfiger, announced a collaboration with OpenAI on Tuesday that will see them co-creating custom AI capabilities in areas such as design, demand planning and customer engagement. It’s the first such fashion deal OpenAI has struck that I’m aware of and will be one to watch.

Also, the heat is rising on fashion’s favourite marketing channel — social media. A landmark trial began in California that will hear arguments on whether social platforms are designed to be addictive and should be liable for resulting injuries. It’s the first in a series of lawsuits brought by hundreds of plaintiffs, school districts and dozens of state attorneys general likening tech companies to Big Tobacco and alleging they’re harming young people’s mental health.

Snap settled in the first case last week, and TikTok joined it on Tuesday, leaving Meta and YouTube as defendants. If companies decide to fight out these suits, they’ll be engaging in very public battles that could put their executives on the stand.

Regulators continue lining up against social networks, too. Lawmakers in France have advanced a bill that would ban the technology for users under 15. The UK also launched a consultation last week it said would “look at options including raising the digital age of consent, implementing phone curfews to avoid excessive use, and restricting potentially addictive design features such as ‘streaks’ and ‘infinite scrolling.’”

Australia took a big step recently by banning kids under 16 from using the technology, and it probably won’t be the last country to do so.

Let’s get to the rest of the news.

Will TikTok’s Algorithm Change Under Its New Owners?

Retraining TikTok: Last Friday, after years of drama, we finally got a TikTok deal. The app’s Chinese creator, ByteDance, spun off the US operations under a new joint venture owned mostly by non-Chinese investors. One thing it didn’t gain ownership of, however, is TikTok’s secret sauce: its algorithm.

Under the terms of the agreement, it will license the algorithm from ByteDance, and there will be some changes to it. TikTok said the new joint venture will “retrain, test, and update the content recommendation algorithm on US user data” in a press release.

What does that mean: There’s no real clarity yet on what effect that could have on TikTok’s US user experience. Many US users may never notice a difference — unless something goes wrong. A number of people complained on Reddit that their For You Page was suddenly filled with bizarre videos, prompting speculation it may have to do with the new algorithm. When contacted, TikTok pointed to a statement on X saying the issues were due to a power outage at a data centre that caused a “cascading systems failure.”

Beyond immediate hiccups, US audiences could conceivably have less access to global content that previously appeared in and shaped their feeds. On the flip side, the retraining raises questions about how content originating in the US will spread overseas, which could have implications for businesses and influencers.

TikTok pointed out in its press release that US users will still get a global TikTok experience, ensuring US creators “can be discovered and businesses can operate on a global scale.” It added that TikTok global’s US entities “will manage global product interoperability and certain commercial activities, including e-commerce, advertising, and marketing.”

Other concerns: These aren’t the only issues to keep an eye on. Among the three managing investors in the new joint venture, each of which holds a 15 percent stake, is Oracle, the cloud computing and software company controlled by Larry Ellison, an ally of Trump’s. Oracle is responsible for US TikTok’s data security and retraining the algorithm, and critics of the deal worry that, under Ellison, the app could push content more favourable to the president. CBS News has already swung to the right after Ellison and his son acquired its parent company, Paramount, over the summer.

TikTok says it hasn’t begun updating its US algorithm yet, but users haven’t wasted time accusing the app of suppressing videos about ICE’s killing of Alex Pretti in Minneapolis, an incident that has ignited outrage across the US. Singer Billie Eilish weighed in, posting on Instagram Stories that “tiktok is silencing people btw.” Other users say posts critical of president Donald Trump are not being seen. California governor Gavin Newsom said on X he is launching a review into whether TikTok is censoring content critical of Trump in violation of California law.

(While it’s not algorithm-related, users have also been deleting the app at higher rates after noticing its privacy policy said it may collect information like race or ethnic origin, sexual orientation and citizenship or immigration status. Some also say they can’t send the word “Epstein” in direct messages.)

In short, TikTok is back in the spotlight for a different reason now: It’s becoming a new political battleground, bringing new considerations for brands advertising on it.

Nobody Knows Who’s Getting Returns on Their AI Investments

The latest survey: PwC released its annual global CEO survey last week, and among the more notable findings was that most CEOs say their companies aren’t getting any financial returns from AI investments. Here’s the breakdown:

• 12 percent saw both increased revenue and decreased costs in the past 12 months
• A total of 30 percent said they’d seen revenue increase (though a portion of those saw costs rise, too)
• 26 percent saw costs fall (but in that group, 1 percent said revenue decreased)
• 56 percent said they’d seen neither revenue nor cost benefits

It could be worse. Remember that study by an MIT-linked group that said 95 percent of organisations were getting zero return on their investments in generative AI?

But wait: Critics of that MIT study were quick to point out issues, like the fact that the authors never said how they calculated that 95-percent figure. (The authors said it was directionally accurate in the study.)

Also, when researchers at the Wharton School at University of Pennsylvania looked at the subject last year, they came to a different conclusion: Nearly three in four business leaders reported a positive return on investment.

Last week, The Wall Street Journal examined a related benefit: CEOs said it was making them more efficient. Then again, that same story found non-management employees said they were saving little or no time at all. Other work has similarly found developers work slower when using AI.

What gives: All the differing results point to the challenge of getting answers from these sorts of surveys. They often rely on self-reporting, have different methodologies and they’re asking questions about a fast-evolving technology that companies aren’t implementing in any uniform manner and are still largely figuring out. The conclusions may say more about the difficulties of integrating new technology into established workflows than they do about the value of the latest AI at this stage.

It’s likely to take some time before we have a clear understanding of how AI is really affecting productivity and what returns it is or isn’t providing. Even then, the results are going to vary by business.

Morgan Stanley, for one, feels optimistic. In a list of 10 predictions for 2026 it released this week, it said concerns over AI adoption rates in the first half of the year would shift to bullishness “on the growing magnitude of AI adoption benefits, as the non-linear increases in AI capabilities are felt.”

Creatives Take Opposing Stances on AI

Pro: Isamaya Ffrench, the prominent makeup artist and founder of Isamaya Beauty, shared her views on AI in an op-ed for The Business of Beauty last week. Creatives shouldn’t fear AI, she said, noting the technology has only strengthened her creative impulse. “The possibilities feel endless and transformative,” she wrote.

Image-maker Nick Knight and stylist Simon Foxton have similarly said AI enhances their creativity. Designers such as Norma Kamali and Collina Strada’s Hillary Taymour have embraced it as well.

Con: At the same time, the technology continues to receive plenty of backlash. On Instagram, many of Ffrench’s followers voiced their concerns that AI will eliminate the jobs of creatives while adding to the world’s already significant environmental problems.

Hundreds of actors, musicians and writers from Cate Blanchett and Questlove to Scarlett Johansson and Roxane Gay have also come out in support of a new anti-AI push by a group called the Human Artistry Campaign. The group doesn’t object to AI on principle but rather to the way tech companies have trained their models on creatives’ work “without authorisation or regard for copyright law.” It says tech companies should sign licensing agreements and strike partnerships to get the material they need.

Ways forward: It’s hard to argue AI developers considered the rights of creatives in the way they trained their models, which basically involved ingesting any data they could get their hands on. Whether those creatives have any recourse will come down to the courts. So far court decisions have been mixed, but this year a number of AI copyright cases are set to take place involving companies like Google and Anthropic that will more clearly define what tech companies do — and don’t — owe the creators of the work they use.

In the meantime, businesses from Disney to Universal Music Group are making deals with AI firms, and more companies are using AI creatively behind the scenes — see above-mentioned collaboration between PVH and OpenAI. I can picture most, if not all, big fashion companies using AI for design in the future, even if they don’t talk about it publicly.

How upset consumers get may depend on whether they notice. Stories keep surfacing about AI-generated songs rising in the charts, for instance. Often it seems listeners have no idea they’re AI. Or maybe they do and just don’t care. In a recent survey, Morgan Stanley found more than 50 percent of listeners ages 18 to 44 said they listened to as much as three hours of AI music per week, mostly on YouTube and TikTok.

Want to dive deeper into an insight from this article?Check out The Brain of Fashion, BoF’s new generative AI tool where you can unlock BoF’s technology archive with a single question.