Controversial Product Listings Continue to Plague Amazon

In what seems to be a defining cultural moment, as brands, including PepsiCo and the NFL, assess and revamp mascots or names deemed racially or culturally offensive, platforms such as Amazon and Facebook are under pressure to proactively address hateful content. Amazon recently announced a $10 million donation to organizations “working to bring about social justice for and improve the lives of Black and African Americans." Amazon announced the donation as brands across America began responding to growing consumer support for the Black Lives Matter movement. But Amazon struggles with a conundrum: Independent, third-party sellers frequently attempt, and succeed at times, to list products that appeal to white supremacists on its platform. In response to questions about controversial content, Amazon previously told Adweek it has comprehensive policies that outline what products may be sold. Amazon uses “sophisticated, automated tools that use machine learning to scan listings and automatically block or remove listings found to be in violation of the policies,” a spokesperson said. That is supplemented by teams of investigators that review the listings regularly. While Amazon declined to specify the precise tools it uses, SEO and digital marketing consultant Joe Youngblood pointed to Amazon’s image and video analysis technology, Rekognition, which the platform itself says “can identify objects, people, text, scenes and activities in images and videos, as well as detect any inappropriate content.” Amazon has pretrained algorithms for Rekognition customers, Youngblood noted, including one called “Detecting Unsafe Content,” which lists online marketplaces as a use case. “There’s ample evidence that Amazon not only has the technology to find products with hate symbols automatically, but that they are a leader in that field,” Youngblood said. And yet, hate symbols remain listed on Amazon. It’s a seemingly intractable challenge. Here’s why. Amazon is really big, and growing For starters, Amazon is really big. Really big. According to ecommerce intelligence firm Marketplace Pulse, Amazon has more than 500 million listings and 2.2 million independent sellers worldwide—and adds 3,100 sellers each day. That’s a lot of products and sellers to monitor. “I think something fascinating with Amazon is something like [a shirt with an oath of loyalty to Nazi leader Adolf Hitler] exists in the furthest corners of the Amazon universe, but unless you’re looking for it, you won’t see it,” said Chris Ross, vp, analyst at research and advisory firm Gartner Ross. “In the Amazon universe, outside of Nazi shirts, there’s probably hundreds of objectionable categories. If you look on the fringes, there’s probably all kinds of stuff considered objectionable or sketchy by someone.” Watchlist of symbols is growing Amazon’s Offensive and Controversial Materials policy and monitoring efforts are hampered by a watch list of hate symbols that is long and complicated. Even powerful algorithms require human intervention. The anti-hate nonprofit Anti-Defamation League (ADL) is lending a hand to help flag extremist content on Amazon. However, listings for problematic products continue to surface. In a recent tally, Adweek found more than 200 listings with symbols the ADL identifies as hate symbols, including the Nazi eagle, noose, swastika and White Lives Matter messaging. We found an additional 2,000 listings with symbols the ADL database says could appeal to white supremacists but are also used by other groups without racist agendas. That 2,200 figure is a tiny fraction—roughly 0.00044%—of Amazon’s overall marketplace. When it comes to eliminating these listings, Amazon faces a challenge similar to the problem of preventing knockoff products and unauthorized sellers, which brands say they face in a never-ending cycle of Whac-A-Mole. In one recent case, director Billy Corben tweeted a link on July 4 to a listing for a face mask with the name Trump depicted as a swastika. The tweet got some additional traction from actor Bradley Whitford. Within about four hours, the listing was removed. https://twitter.com/BillyCorben/status/1279577554089803776 In another case, it didn’t happen so quickly. In May 2019, Adweek reported a Prime-eligible neo-Nazi flag to Amazon. The flag features the Othala rune, which is reportedly replacing the swastika as the primary symbol of neo-Nazis and has been burned at rallies alongside crosses, and a Sonnenrad, or black sun, which has also been adopted by Nazis, and later neo-Nazis. The flag remained on the site until just prior to the publication of a July 25 BBC investigation. Obscure and ambiguous hate symbols Further complicating matters is the ever-growing number of hate symbols, especially ones with dual meaning. The ADL’s online database includes 214 listings. Some are well-known. Many, such as the logo for the white supremacist band Skrewdriver or the Totenkopf or Wolfsangel symbols, are not. Additionally, 24 symbols include disclaimers that they must be considered in context because they are also used by nonextremists. Take, for example, the life rune. Martin Findell, assistant professor of historical linguistics at the University of Nottingham, said this rune represents the sound Z in the oldest form of the runic alphabet, and most modern-day pagans interpret the symbol to mean “elk” or “protection.” However, it was also the emblem of the Nazi project Lebensborn, which was “engaged in actively trying to breed a superior race,” and today, Findell said, the symbol remains “very popular with the far right.” A recent search surfaced about 60 Amazon listings with the life rune. But, as a result of that dual appeal, Findell said, whatever listings Amazon carries with this rune depend on the context in which they are being sold and what they mean to the customers who buy them. Consumer reaction Other online retailers and platforms, such as eBay and Google, face similar challenges, though perhaps not at Amazon’s scale. As platforms, they take a hands-off approach, which allows them to distance themselves from controversial content, said Chris Ross, vp, analyst at research and advisory firm Gartner. Users make a mental distinction between the objectionable content from third parties and the values of the platforms themselves. Part of the reason consumers are so forgiving is because Amazon removes the products that make headlines, said Elizabeth Marsten, senior director of strategic marketplace services at performance marketing agency Tinuiti. Amazon did not respond to questions about how it decides to remove listings. Historically, Amazon has removed objectionable listings with no or limited comment.