The Current

Facing disturbing content daily, online moderators in Africa want better protections and a fair wage

Content moderators in Kenya say social media networks have been exploiting them for years, offering low wages and inadequate mental health support.

Kenya-based moderators for social media and A.I. platforms paid $2 Cdn an hour: former worker

A crowd stands in front of an office building. They are waving papers in the air
Content moderators gathered outside Sama AI offices in Nairobi demanding the company pay them their owed wages. (Submitted by Trevin Brownie)

Read transcribed audio

Trevin Brownie had to sift through lots of disturbing content for the three years he worked as an online content moderator in Nairobi, Kenya.

"We take off any form of abusive content that violates policies such as bullying and harassment or hate speech or violent graphic content suicides," Brownie told The Current guest host Mark Kelley.

Brownie has encountered content ranging from child pornography, material circulated by organized crime groups and terrorists, and images taken from war zones around the world, including Russia's invasion of Ukraine.

"I've seen more than 500 beheadings on a monthly basis," he said.

Brownie moved from South Africa, where he previously worked at a call center, to Nairobi, where he worked as a subcontractor for Facebook's main moderation hub in East Africa, which was operated by a U.S.-based company called Sama AI.

App icons for Facebook, Whatsapp and Instagram are pictured on a shattered glass screen.
Facebook's parent company Meta no longer uses Sama as an outsourcing partner. (Dado Ruvic/Reuters)

Despite the mentally straining nature of the job, content moderators working in Kenya say Sama AI and other third-party outsourcing companies took advantage of them. They allege they received low-paying wages and inadequate mental health support compared to their counterparts overseas.

Brownie says the nature of his job severely damaged his mental health. PTSD has become a common side effect he and others in this industry now live with, he said.

"It's really traumatic. Disturbing, especially for the suicide videos," he said.

"Watching those has certainly made most of us, if not all of us, senseless when it comes to emotions or feelings towards death."

Facebook's parent company Meta no longer uses Sama as an outsourcing partner. However, the two companies are currently facing multiple lawsuits in Kenya. The allegations range from porous working conditions, unlawful dismissal following attempts to unionize, blacklisting workers, and withholdment of pay for laid-off staff.

Brownie, one of the 184 petitioners in one lawsuit, says he's among the former employees Sama AI laid off without paying wages they are owed.

Meta unsuccessfully attempted to challenge its involvement in the case, arguing that the company had no official presence in the country, unlike Sama AI.

It's not the only lawsuit related to content moderation under way in Kenya.

Majorel, another content moderation company that operates in Africa but is based in Luxembourg, is being sued for discriminatory hiring practices for allegedly blacklisting Sama employees following their dismal.

In a statement, Sama refuted the moderators' allegations of union-busting, and underpaying their staff.

"We recognize that content moderation is a tough job and have paid wages that are 4x the minimum wage and 2.5x the living wage in Kenya as a recognition of the challenges of that work," it said, adding that it paid higher wages for expats living in the country.

A man stands in front of a grey curtain, looking to the camera with his arms crossed.
Trevin Brownie is among 184 petitioners in a lawsuit against Sama AI and META. (Submitted by Trevin Brownie)

But Brownie says that's far from reality. He says concerns around work conditions and pay were frequently brushed aside. 

"What they tried to enforce is that we were one of the 10 highest paying jobs in the country," he said.

Brownie says he earned the equivalent of $2 Cdn an hour for his work. After taxes, he took home approximately $500 per month. Meanwhile, he says people doing the same kind of moderation work overseas can make $20 an hour.

"It's just enough to keep me living paycheque to paycheque. But because we need these jobs and our families back home rely on the support we do financially, we stay and work," he said.

Real people are behind AI, too: expert 

Earlier in May, more than 150 content moderators who work with the artificial intelligence (AI) systems used by Facebook, TikTok and ChatGPT, from all parts of the continent, gathered in Kenya to form the African Content Moderator's Union. It claims to be the first such union in the continent.

The union is calling on companies in the industry to increase salaries, provide access to onsite psychiatrists, and a redrawing of policies to protect employees from exploitative labour practices. 

In a statement, Sama said it applauded the union's formation. "Sama has policies and training in place to support freedom of association, and if/when that occurs, Sama will happily engage with that group," the company said, adding: "We care deeply about the health and emotional well-being of our team."

For Sonam Jindal, stories like Brownie's are all too common, and holding employers accountable for actions in foreign jurisdictions is often challenging. 

"The part of what makes this industry a little bit complex and difficult is because there are so many different actors involved and it kind of leads to a situation where there's limited transparency around who is actually accountable at a given point," Jindal, program and research lead on AI, labour, and economy at the non-profit group Partnership on AI, told Kelley.

"At the moment, there is not a whole lot of regulation," Jindal said.

Jindal says a key obstacle to getting better protections for content moderators lies in how people think social media platforms work — and now, she said, that problem extends to new and growing artificial intelligence (AI) platforms.

WATCH | Artificial intelligence makes opening statement at U.S. Senate hearing:

Artificial intelligence makes opening statement at U.S. Senate hearing

2 years ago
Duration 1:13
To open the hearing on artificial intelligence, Democratic Sen. Richard Blumenthal played a recorded statement created entirely by OpenAI's ChatGPT and AI voice cloning software trained on his own speeches to mimic his voice. OpenAI CEO Sam Altman also testified at the hearing, calling on the government to regulate artificial intelligence.

ChatGPT, for example, has gained popularity since its launch last year, said Jindal.

"I think a lot of the narrative is around, 'A.I. is so great.' We've made these technological advances … but those advances are based off of human labour and input. And I think we need to recognize how important they are." 

AI platforms like ChapGPT need people like Brownie to classify and filter data, Jindal explained, checking for inaccuracies and removing harmful content, allowing them to operate smoothly.

"There is no social media platform if there are no moderators to make sure that the material displayed on there is the actual material," Brownie said. 

As ChatGPT and other AI platforms grow in popularity, Jindal says a significant overhaul in how industry leaders carry out business both abroad and domestically is needed. 

Partnership on AI published a five-point list of recommended changes, which includes paying workers a living wage according to their location and establishing clear and regular communication mechanisms between workers and managers.

Jindal says lawmakers in all jurisdictions also have a key role to play in implementing policies that protect workers from exploitative practices.

She also wants to see the UN's International Labour Organization involved in oversight. 

In the meantime, Brownie says his time in Kenya will come to an end after litigations conclude.

"I'm going to go back to [South Africa] to try and rebuild my life, find my sanity, and just start over," he said.

ABOUT THE AUTHOR

Niza Nondo

Journalist

Niza Nondo is an associate producer for CBC Radio's The Current. He's particularly interested in telling stories about the Middle East and Africa. Niza previously worked at Global News Toronto and in his home country, Zambia, at ZNBC News.

Add some “good” to your morning and evening.

Get the CBC Radio newsletter. We'll send you a weekly roundup of the best CBC Radio programming every Friday.

...

The next issue of Radio One newsletter will soon be in your inbox.

Discover all CBC newsletters in the Subscription Centre.opens new window

This site is protected by reCAPTCHA and the Google Privacy Policy and Google Terms of Service apply.