For an actress forever associated with the line “I want to be alone,” Greta Garbo is suddenly keeping some unusual company. Thirty-six years after her death, she is the face of a campaign for SKF, the Swedish maker of bearings, recreated with AI to help sell machinery that most of us will never see. My first reaction was not alarm. It was that the idea made sense. There was a real connection to the company’s history, a reason for revisiting it now, and the involvement of her family. I could see what the brand was trying to accomplish.
That last part is worth dwelling on, because SKF has a problem that sneaker companies and fast-food chains do not. Bearings are a little like good referees: essential to the game, and noticed only when something goes wrong. An industrial technology business has a very different relationship with the general public than a consumer brand does. Nobody wears an SKF hoodie. Getting people to talk about SKF at all, including people who will never buy a bearing but might work for, invest in, or recommend the company, has potential value of its own.
A campaign built on two returns
SKF’s explanation of the campaign rests on two returns. The first is Garbo’s. Before she was a star, a teenage Greta Gustafsson appeared in an industrial promotional film that featured SKF. The second is a financing idea. In 1919, SKF let customers pay for an investment through the energy savings it produced, and the company is reviving that principle for its magnetic bearings, which suspend a rotating shaft without physical contact. SKF says this reduces energy consumption, wear and maintenance, and it frames the stakes broadly, claiming that roughly a fifth of the world’s energy goes to overcoming friction. It also says it developed the initiative with Garbo’s family and Harriet Brown & Company, which helps manage her estate.
Put together, the creative idea ties the company’s past to a current offer, which is a reasonable premise for an advertisement. A brand reaching back a century to find a star who once appeared in its own promotional film is not borrowing fame from a stranger. It is, in a sense, calling an old colleague.
“By bringing Greta back, we hope to open more eyes to frictionless and energy-efficient technology.”
Per Nilsson, SKF Global Head of Communication and Brand, campaign announcement
It is also a more indirect value proposition than simply cutting a price. A savings-funded investment asks the customer to understand how the savings are measured, how the financing works and why the technology can be trusted to deliver. Garbo cannot do that work on her own. A glamorous face can open the door, but an engineer with a spreadsheet still has to walk through it, and the product has to perform once it is installed.
I also would not assume that many of today’s industrial buyers feel a strong personal connection to Garbo. Some people in advertising may appreciate the historical reference more than the intended customer does, and there were certainly other creative approaches that could have communicated the offer well. But being the only possible way to tell a story is an unreasonable standard for an advertisement. Almost no campaign clears it. What this execution gave SKF was a second story, about the way the campaign itself was made, and that, in my view, is part of its purpose.
Attention is not the same as proof
The reaction has included criticism. Peter Bradshaw dismissed the ad as dull in The Guardian, and he is entitled to dislike it. A critic’s judgment about the work is worth hearing. It is also one review, and on its own it tells us little about how customers, performers or the broader public have responded.
My own read is that this is a successful awareness idea. SKF has prompted coverage and a conversation that extends well beyond the mechanics of its product. I want to be clear about what kind of claim that is. It is an assessment, not a measurement. Reach, sales and return on investment are separate questions, and answering them would take evidence that nobody outside the company has yet.
Campaigns can also serve an experimental purpose. A company may want to show what is possible, learn an unfamiliar technique before its competitors do or establish itself as the kind of organization willing to try something new. For an engineering brand, a reputation for innovation is not a vanity metric. There can be real value in doing that well, even before anyone can attach a sales figure to it.
What I learned from a digital Sam Walton
My openness to this kind of work comes partly from an experience I had in Bentonville. While visiting Walmart headquarters for a meeting, I stopped by the Walmart Museum and encountered its virtual Sam Walton. A moderator helped guide the questions, and Walton’s digital recreation answered with stories and views about business. I enjoyed it. Hearing those ideas delivered conversationally made them feel personal in a way a plaque on the wall never would. A plaque tells you what someone believed. This felt closer to hearing him explain why.
At the time, I was not sure how it had been built. StoryFile’s account of the project describes an actor providing the body performance, a face and voice recreated from historical material, and the ability to answer several hundred questions. Coverage of the exhibit’s opening describes it drawing on Walton’s autobiography and recordings from the company’s archives. In other words, what charmed me was already AI-assisted reconstruction. Reporting on the Garbo ad describes a similar blend of generative tools and a human actor contributing to the performance.
That distinction matters to me. What moved me in Bentonville was an experience built from someone’s recorded history, but it was still a representation, shaped by choices about which material to use and how to present it. I can imagine considerable value in developing that idea further. Future generations could explore a person’s documented thinking through questions, stories and context, the way a good biography lets you argue with its subject. For that to be useful, the representation has to stay faithful to what is known and be honest about its limits. A convincing answer invented on someone’s behalf is not insight into that person. It is a guess wearing their face, and it could leave people with a false understanding of who they were.
Permission, representation and work
The same issue applies when a brand recreates a performer. Family involvement matters to me, particularly when the people involved knew the person. Garbo’s great-nephew Craig Reisfield, who knew her, told The Guardian he was impressed by the recreation. I give that perspective real weight. I would also want the people responsible for any legacy to ask whether the message fits the person they remember. Approval, legal rights and faithful representation are related questions, but they are not the same question. A family can authorize a portrayal without proving what the individual would have thought of it, and that seems especially worth remembering with a star who guarded her privacy as carefully as Garbo did.
Then there is the question of work. When people object to AI performances, I want to understand the effect they are describing. A performer worried about losing employment, or about having a voice or likeness imitated without permission, has something concrete at stake, and I give particular weight to those perspectives. How much displacement these tools will cause is not a settled question, and it should not be treated as one. General discomfort with AI deserves discussion too, but it does not answer every question about a particular use.
I regard AI as a tool that creative people should have the opportunity to benefit from. That belief does not make unauthorized impersonation acceptable. People should have a meaningful say in how their identity is used. Sir Michael Caine’s partnership with ElevenLabs, announced in November 2025, illustrates one version of that say. He joined the company’s Iconic Marketplace, through which businesses can request licensed access to participating talent. “It’s not about replacing voices; it’s about amplifying them,” Caine said in the announcement.
I see the appeal of that choice. An authorized voice model can make new work possible without requiring the performer to record every individual reading, and depending on the agreement, it can create another source of income. The terms matter enormously, and one licensing arrangement cannot tell us how an entire profession will fare. Still, it is a practical example of a performer participating in the opportunity rather than watching it happen to him.
A face is a brand, too
This is where the Garbo campaign becomes a branding story rather than just a technology story. A recreation can imitate aspects of her appearance and performance only because there was a distinctive person to draw from. Like any strong brand, a recognizable identity is built over decades, and its value comes from that accumulated meaning, not from the pixels. Brand owners already understand licensing: you can let someone else use your mark, but you set the terms, protect the quality and get paid for the equity you built. Authorized likeness licensing applies the same logic to people.
For some creators, authorized uses could eventually support their families after their deaths. Others might direct that value toward a foundation or a cause they care about. I am describing a possibility, not making a claim about the financial terms of SKF’s campaign, which have not been disclosed.
Consumers have a part in this too. Choosing an authorized version acknowledges the person whose identity gives the work its appeal. A similar-sounding performance is not automatically an infringement, but deliberately seeking out an unauthorized impersonation raises a different question about how much we value the original. Meanwhile, new actors, musicians and artists will keep bringing personalities and ideas that change what audiences care about. Their ability to create something distinct still matters, even as the tools for imitation get better.
That is why I see promise in SKF’s experiment. It did not grab a famous face because it was available. It found a real thread in its own history, worked with the people entrusted with the legacy and attached the result to an offer that still has to prove itself on the factory floor. For any brand tempted to borrow a legend, that is the lesson worth keeping. The technology can extend the commercial life of an identity. The responsibility is to make that extension one the people entrusted with the legacy can stand behind, and one the product can live up to.