TikTok “unemployed guy” was actually AI-generated

Following an investigation by CBC News’ Visual Investigations team, TikTok deleted the “Unemployed flex” account, citing a violation of its community rules. According to the social medium, it was not sufficiently clear that this content had been created by AI.

Josh – this young white man in his twenties, showing his adventures looking for work and complaining about the role of immigrants in the job market – isn’t real. Some TikTok users noticed that the videos were produced by AI, stating so in the comments. However, others were duped into giving advice to the avatar.

Many Internet users also reacted to content they perceived as racist, believing that the person behind the message actually existed. In some cases, Josh – the fictional character – even replied in the comments to defend himself, reinforcing the illusion.

In a video, the young man can be seen complaining about people from India, who he claims have taken every available job, including at Tim Hortons.

In a statement, Tim Hortons said the spread of such videos was frustrating and worrying for the company, adding that it had struggled to get them removed.

In another video, Josh criticizes Canada’s immigration policy, wondering why so many people are admitted to the country when there aren’t enough jobs for everyone. “I can’t believe I lost the job to someone who barely speaks English,” adds the AI character.

The company behind this Tik Tok account is Nexa, which develops software used by other companies to recruit staff. In Josh’s videos, the Nexa logo appears in the background. The company’s founder and CEO, Divy Nayyar, said in an interview with CBC News that this method is a subconscious advertising placement.

Mr. Nayyar told the media that he wanted to have fun with the idea, shared by some, that Indians are invading the job market. He says he created the character of Josh to “humorously” connect with those inclined to share this opinion: young graduates looking for work.

“This type […] of highly polarizing narrative is typical of far-right groups,” said Markus Giesler, professor of marketing at York University, in an interview with CBC News. “For a company to adopt a campaign tone like this to attract customers is extremely problematic, totally unethical, and unheard of,” he continued.

Nayyar explained that one of the tools his company used to create this content was Google’s Veo. The latest version of the software, Veo3, launched in May, can produce far more realistic videos from simple text than previous versions. Obvious clues such as characters with extra fingers or physical inconsistencies are now rare with Veo3. What’s more, the audio is closer to real human voices, and lip movements are well synchronized.

TikTok has not commented on the controversial message in the videos. The platform only indicated that the “Veo” watermark that was displayed on the bottom right of Josh’s videos, did not comply with its rules. The social medium said AI-generated videos featuring realistic scenes should be flagged with a more explicit label, caption, watermark or sticker.

Nayyar claims that he wanted to produce very realistic content, assuming that the public would use their common sense to guess that it was produced by AI. He claims that TikTok automatically tags this type of content, which is false: tags are not automatic, reports CBC News.