Courtesy of Noah Levenson
The website Stealing Ur Feelings accessed your webcam to make you the star of the story, and it deployed emotion recognition AI, interactively testing your reactions to various content.
In May 2021, a group of journalists at the Financial Times did something I’d never seen before. They took a copyrighted work, changed some elements, and—without permission—rebranded it as their own and published it with the full support of the Financial Times media apparatus.
The work they appropriated happened to be mine—and they did it, in part, by having me sit for an interview for an article that never appeared.
The following is a story from the weird frontier of interactive journalism, where journalists confuse the language of news and entertainment and commit acts of cynicism that we may not yet have the vocabulary to explain. It’s not the worst thing that’s ever happened to me. But it might be the strangest.
In 2018, the Mozilla Foundation awarded me a grant to produce a work of creative technology in the public interest. The result was Stealing Ur Feelings, an interactive website about the societal risks posed by emotion recognition AI. Stealing Ur Feelings accessed your webcam to make you the star of the story, and it deployed emotion recognition AI against your face, interactively testing your reactions to various content and revealing how things might go wrong when the algorithms misclassify you.
What does “inspired by” mean in the context of journalism?
The project was successful. It premiered at the Tribeca Film Festival’s interactive competition and toured a number of documentary festivals worldwide; it won the 2020 Webby Award in the Interactive and Mixed Reality category; and it made the front page of Y Combinator’s Hacker News, racking up 104 comments from tech folks debating the outcomes of emotion detection technology. During its run, Stealing Ur Feelings received substantial international media coverage, including articles in Scientific American, Engadget, CBC News, El País, and the Museum of the Moving Image’s Sloan Science & Film publication. As the creator of Stealing Ur Feelings, I provided a number of interviews to journalists about the project for both written pieces as well as broadcast segments.
In January 2020, I received an interview request from Aendrew Rininsland, a journalist at the Financial Times. Rininsland’s email to me overflowed with praise for Stealing Ur Feelings. Rininsland, who uses they/them pronouns, said they were “absolutely blown away” by the work and that it was “one of the most effective pieces of visual storytelling I’ve seen in years.” They continued: “I particularly liked how stock imagery is used to record topical user sentiment that is then later reported back to them; absolutely brilliant.”
Little did I know that Rininsland liked my work so much that they would go on to do a new version of it.
Rininsland’s email said they were “researching some of the issues you brought up for a Tech desk story and minimalist visual project” and asked whether I would have time for an interview. I agreed, and we set a date.
In our conversation, conducted via Google Hangouts, Rininsland continued to praise my work, explaining that Stealing Ur Feelings was to feature heavily in the story they were working on. They asked many of the same kinds of questions that other journalists asked: Why did I become interested in this topic? How did I discover the patents featured in the story? Could I point to the supporting materials referenced in the video? They also asked a number of questions related to the specific software libraries used to engineer the project. This was peculiar in the context of other interviews I had sat for; journalists covering the topic tended to focus on why I built Stealing Ur Feelings, not how I built it.
In a follow-up email after my interview, I asked Rininsland to let me know how the piece develops. “Will let you know when we’re a bit further along with it,” came the reply. But the promised piece never materialized, and I never thought about it again.
That is, until May 11, 2021—roughly a year and a half later. Rininsland had reappeared in my in-box: “Just a head’s up, my team and I have been working on an interactive piece touching on many of the same topics … It goes without saying that your project was a big influence on the work we did!”
“It’s publishing at around noon London time tomorrow,” they wrote.
The next day, Rininsland followed up with a link. It was a tweet from the Financial Times to their seven million followers: “How badly wrong can things go when companies use AI to identify your emotions? Play this FT game to find out.”
I don’t need to convince you of all the ways in which they took the premise, mechanics, distinctive features, and details of Stealing Ur Feelings—you can decide for yourself. Watch https://stealingurfeelin.gs, then head over to https://ig.ft.com/emotion-recognition.
The strangest part, to me, is why they would copy even the arbitrary aesthetic choices. For no good reason, Stealing Ur Feelings punctuates segments by raining down a shower of bouncing emojis. Rininsland’s version does the same thing. Where Stealing Ur Feelings tests your reaction to stock footage of dogs, Rininsland’s version shows you cats. At the conclusion of Stealing Ur Feelings, the viewer is presented with a selfie revealing their “stats”—the quantitative emotional data collected by the algorithms during the experience. Rininsland’s version for the Financial Times does that too.
Courtesy of Noah Levenson
On Twitter, Financial Times staffers, mostly verified users, were congratulating each other for their vision and innovative spirit. Said Joanna S. Kao, data journalism tech lead: “Together, [Rininsland, Madhumita Murgia, and India Ross] started to develop a story concept … This project couldn’t have existed without every single person.” She continued: “Creating new things isn’t straightforward, and it carries risk. It requires a different way of working, collaborating, and publishing, but it’s exciting to see when investments in experimentation pay off.” Neither I nor Stealing Ur Feelings was mentioned.
Martin Stabe, a data editor, tweeted: “The amazing project was conceived by [Rininsland] and brought together with in a vast effort with [Emma Lewis, Caroline Nevitt, Aleksandra Wisniewska, India Ross, and Madhumita Murgia].” Amid the celebratory jubilation, I found one lone tweet with a link to Stealing Ur Feelings, acknowledging the derivation—perhaps unsurprisingly, from the Financial Times’ assistant general counsel, John Halton: “See also this similar exercise (acknowledged in the credits for the FT interactive).”
If you can find the credits, you’ll see the fine print he’s referring to: “Project inspired by stealingurfeelin.gs by Noah Levenson.”
This begs the question: What does “inspired by” mean in the context of journalism? And if the Financial Times isn’t producing journalism, then what, exactly, do they claim to produce?
[Editor’s Note: The FT responded as follows to a question about specific details in this article and whether it had given appropriate credit to Noah Levenson: “We often cite other work in the reporting we undertake and in this instance an appropriate credit was given to Noah Levenson at the end of the feature.”]
The Financial Times never received my permission to appropriate Stealing Ur Feelings. While Rininsland and their team were paid as employees of the Financial Times, I received no compensation. The interview I provided to Rininsland was under the guise of their reporting on Stealing Ur Feelings—not creating a new version of it to monetize for ad revenue. And while I conceived of and engineered Stealing Ur Feelings, I was hardly the only person responsible for producing it. Stealing Ur Feelings required the contributions of many artists, writers, and technologists, all of whom worked for low or no pay, as Stealing Ur Feelings was a nonprofit enterprise.
In the arts and entertainment, there exists the concept of an homage. Films borrow ideas about framing and editing from other films. Musicians steal riffs from each other all the time. We’re familiar with cases where homage has crossed the line into copyright violation, and litigation ensues: Robin Thicke had to pay the estate of Marvin Gaye for “Blurred Lines,” for example.
But is the Financial Times a producer of entertainment? When Rininsland requested an interview with me, they proclaimed themself a fan of the work. The premise of the interview was that they were going to report on Stealing Ur Feelings. What business are journalists in when instead of talking about a work or critiquing it, they appropriate the concept and release it as their own invention?
At the time of Rininsland’s interview request, Stealing Ur Feelings had screened at Open City Documentary Festival in London, Montreal International Documentary Festival, and Camden International Film Festival. It had exhibited at the Tate Modern in London and at the Glass Room gallery in San Francisco. I’m not famous, but the surreality of this event is brought into relief if you imagine that I was. It’s as if a reporter interviewed a little-known director about her new movie—and then, instead of publishing the interview, the reporter just decided to film a movie exactly like it and release that instead.
Perhaps that isn’t exactly illegal. But it’s sleazy.
My agreement was based largely on society’s shared understanding of what the product we call “journalism” is.
In Rininsland’s emailed interview request, they identified themself as a “visual journalist.” My trust in the professionalism of journalists—and the kinds of ethical standards outlined in the FT’s own Editorial Code of Practice document—is the reason I agreed to be interviewed.
My agreement to the interview was also based largely on society’s shared understanding of what the product we call “journalism” is. When journalists ask to interview you about your art project, it’s unfathomable that they might secretly harbor ambitions to enrich themselves by copying it. Had Rininsland instead identified themself as a “web designer” or “creative director,” it would have been silly for them to use the word “interview” to describe the request. It would have also suggested a different set of assumptions: A creative director who designs websites for the Financial Times might not be acquainted with or held accountable to the ethical standards put forth in the FT’s Editorial Code of Practice. But I assumed—perhaps naïvely—that anyone claiming to be employed as a journalist must do so.
The landing page at https://ig.ft.com bears the heading “Visual journalism.” Along with their clone of Stealing Ur Feelings are pieces detailing the latest data on global COVID-19 infections, U.K. politics opinion polls about the Scottish National Party, and recent ideas about climate change. Some are behind a paywall, with digital subscriptions advertised as costing $40 per month. Many are credited to the “FT Visual & Data Journalism team” and the people I saw mentioned in tweets.
If you explore this section of the Financial Times, you’ll discover that these pieces of “visual journalism” seem to have little in common with each other. Besides their “new” interactive piece about emotion detection AI, there are data visualizations, interactive quizzes, and annotated illustrations. What coheres these disparate products is one broad idea: They’re all attempts to tell stories in new ways using the unique medium that is the web browser.
With legacy journalism, grifts are harder to disguise. We expect the credits for a written piece to appear prominently under its title, not buried in a link in a hidden navigation tab. Sources are attributed using quotation marks. It’s comical to imagine an article in The New York Times or The Wall Street Journal purporting to be original news while having the words “inspired by” under the headline.
Similar conventions do not exist for journalism that takes the form of a web-based software program. My charitable side suggests that Rininsland and their colleagues at the Financial Times are under too much pressure to bother confronting this idea. My cynical side suggests they knowingly exploit it.
Whichever it is, this case study certainly suggests that the product exported by the Financial Times is something much different than journalism. What that product is, precisely, remains unclear. In appropriating Stealing Ur Feelings, the FT was acting like a stock music house that produces soundalikes of pop hits to use in TV commercials. Less charitably, it could be compared to a site that uses machine learning to rephrase scraped articles to sell ads against.
In FT staffer Joanna Kao’s effusive Twitter thread celebrating the project, she said: “There was a lot of knowledge gained during this project. I know that at least [Rininsland] and [Emma Lewis] likely have enough journalism/tech talks to give about this to last them awhile.”
In this case, I have to take her word for it. What worries me is that they learned the wrong lesson.
