SerpApi.Org
Social-media users are starting to sense a trick behind every product review or fun-spirited trend—and they aren’t just being paranoid.
You could say just two simple words, “Wow, OK,” and convey dozens of different messages with tone alone. The way people say things is sometimes much more important than the actual words coming out of their mouths. This idea is exactly what spurred the online acting challenge and trend of saying the same phrase four different ways—usually supportive, disappointed, sarcastic, and flirty. The format became popular in June 2026 on TikTok and Instagram with thousands of videos under different iterative hashtags like #wowokay and #samephrase.
As the trend grew, content creators began using different phrases and trying it out with friends, each taking their turn to say the phrase in four different emotions. It seemed like a totally benign trend that let people test out their acting chops—but not everyone thought so.
Clara Fulks, a digital ethicist and the CEO of tech consulting firm North Star Strategies, posted an Instagram reel on June 19 theorizing that this trend was actually manufactured for AI data collection. “Manufacturing trends is an easy win if you’re a big tech company in a data drought,” she wrote in the caption.
Fulks, who holds a master’s degree in speech and language processing, explained that large language models (LLMs) are still bad at emotional recognition and a trend like this with a simple, consistent format will provide troves of human data.
Olga Kokhan, CEO and founder of data services company Tinkogroup, explained to Cybernews, “Thousands of users saying the same phrase with different emotional expressions create a dataset that teaches AI systems to distinguish between what is being said and how it is being said.”
Fulks went on to say that Meta uses data from user-provided content like public Instagram reels to train its AI models, so this trend (whether manufactured or not) will help AI emotional recognition improve. Other viral trends, like “hugging my younger self” or “2016 again,” also drew concerns that users were helping AI learn how people age by providing pictures of themselves, either directly to chatbots or on social media, over the course of many years.
These suspicions about trends helping train AI become understandable when online marketing companies such as Doublespeed use AI models to generate ads posing as actual human recommendations.
Doublespeed has made an AI-character posing as a 62-year-old mom from Phoenix that could, for example, advertise workout products that helped fix her back pain. To an average viewer, it might be unclear this is an ad and even more unclear that this person doesn’t even exist. These characters’ entire purpose is to seem like real people in order to sell products.
Some are undisclosed ads, meaning they do not contain any language or indication they are paid promotion, which is already a huge problem with human content creators. But even if a Doublespeed-run account tags the content as sponsored, it’s still against TikTok and other platforms’ policies because of its AI use and phone farming, according to reporting from New York Magazine.
Backed by venture capital firm a16z, the company’s website claims clients will “Never pay a human again” to create ad content. These fake content creators could be the exact types of AI benefiting from complex emotional data provided by real creators. And it is taken with the explict intention of making human creators obsolete, as Doublespeed CEO and co-founder Zuhair Lakhani said on X.
Even beyond AI-related issues, other concerns that trends are carefully planned marketing schemes have come to light as well. Indie-rock band Geese had a sharp rise in popularity in mid-2025, with many fans evangelizing the band’s unique sound across social media. The band formed over a decade ago among high school friends, but searches for the band on Google rose over 1,400 percent in the past five years, peaking at the end of 2025.
In April 2026, writer and musician Eliza McLamb reported on her Substack that music marketing company Chaotic Good was running what it called “narrative campaigns,” which involved the creation of seemingly independent social-media pages and employees posing as genuine fans in videos, comments, and posts. Some fans and onlookers were thrown, calling the band a “psy-op” and feeling tricked into trusting online hype as organic rather than paid and calculated.
Wired reported on the controversy, noting “these tactics follow a golden rule of so much online discourse, which another of Chaotic Good’s founding partners summed up in that Billboard interview: ‘Everything on the internet is fake.’”
That fakeness feels increasingly palpable for users today, who may now sense a trick behind every genuine product review or fun-spirited trend. Instead of a creative outlet or place to learn, the internet seems like a marketplace where data is bought and sold for the purposes of training AI and surreptitiously selling products.
This post originally appeared at inc.com.
“Click here to subscribe to the Inc. newsletter: inc.com/newsletters"
