Google Drops Firm Reviewing YouTube Videos

Google ended its contract with ZeroChaos, which supplied human ad raters, leaving many unemployed.
Image may contain Human Person and Machine
John Holcroft/Getty Images

Earlier this year, after multiple reports that ads appeared with YouTube videos promoting hate and terrorism, several big brands, including Walmart, PepsiCo, and Verizon quit the video company's ad network. In response, YouTube parent Google increased both the number of "ads quality raters"—the people who spend hours watching YouTube looking for violence, profanity, even death—and the volume of videos those raters viewed.

These raters are temporary workers employed by companies contracted by Google, and Google recently dropped one of its larger contractors, ZeroChaos. “Effective 7/31/2017 9pm PST, your current assignment with ZeroChaos will come to an end,” an email sent to employees reads. “Thank you for all your hard work in connection with the Google Quality Rating Project. Unfortunately, Google has terminated the Quality Rating Project that ZeroChaos was managing, and as a result no further opportunities are available. We appreciate your commitment and dedication.”

The workers' ratings serve two purposes: they classify inappropriate content, and those classifications help educate Google’s AI, a technology that will eventually flag sensitive or offensive content on its own, without human help.

It’s unclear whether Google has moved more of the rating work to other contractors or its AI. In a blog post Tuesday, Google said that “over 75 percent of the videos we’ve removed for violent extremism over the past month were taken down before receiving a single human flag.”

Google doesn’t say how many temporary workers it employs for its ads rating efforts (for “confidentiality” reasons, according to a spokesperson), but the number could exceed 10,000, according to one estimate. Other contractors include Leapforce, Lionbridge, and Appen, but comments on discussion boards used by workers suggest ZeroChaos paid higher wages.

This kind of work tends to attract people who prefer more flexible working conditions, such as workers near retirement age, stay-at-home parents, and individuals with physical disabilities, many of whom have limited employment opportunities.

“I have two autistic kids and although I have several eggs, this was my biggest chunk of money, and I need it,” one ad rater said in a forum.

“I am at the end of my rope. I don’t know what I am going to do for money since my partner is making above the threshold for unemployment. I can’t believe this,” said another. “I don’t know how we are going to find work again, because if we all lost our job today, what are we going to do about oversaturating the other companies?”

As recently as May, ZeroChaos told employees their contracts could continue into 2019, according to an email reviewed by WIRED. Some who were training for the program, meanwhile, waited for months to start working at ZeroChaos, only to be told there would be no work for them. “The company led me on for six months, only to tell me that Google pulled the contract,” says one ad rater, who showed WIRED a February email that said he had been accepted to the program, and seven subsequent emails, each delaying his start date. “The loss of jobs for such a large number of people—particularly by a company that brags about its liberal ideology, ‘Don’t be Evil’—deserves public attention,” says another.

“We’re working to ensure we have the right vendor partners and processes in place as we increase our efforts to stop the abuse of our platforms, and to keep our users safe,” says Ty Sheppard, a Google spokesman. "We work with many other vendor partners in this space and opportunities at these employers will grow." ZeroChaos declined to comment. The company tells workers on their first day that their employment is at-will and ZeroChaos can terminate their contracts.

To some, the abrupt end to the contract highlights the differences between Google’s full-time employees, who have secure and well-paying jobs, and a hidden workforce that makes its products run more smoothly. “For some, tech can be incredibly lucrative and rewarding," says Sarah T. Roberts, a media studies scholar at UCLA who studies commercial content moderation. But, she says, many others "don’t fall into those types of categories, and are more subject to a kind of labor precariousness.”

The annual salary for a software engineer at Google is $127,000, according to the workplace website Glassdoor. Meanwhile, ZeroChaos' ad raters earned $15 per hour for up to 29 hours a week; they did not receive raises or paid time off, and signed a confidentiality agreement that continues after their employment ends.

Tech platforms host enormous amounts of content—YouTube alone sees 400 hours of content uploaded every minute—and more content translates to more revenue for these companies. “These platforms don’t exist for goodwill or the public interest, they are for profit,” says Roberts. For now, these moderators are still necessary, a human filter to watch, assess, and delete the worst of it so others don't have to see it.