Thousands of women working remotely in India are being exposed to a staggeringly high rate of traumatic content to train artificial intelligence systems, according to the Guardian.

What's happening?

Content moderators operate in the background by nature, and the outlet described them as "ghost workers."

Content moderation falls under the broader "trust and safety" field and has long been a controversial function for platforms that host user-generated content.

In 2020, Facebook agreed to a $52 million settlement for "current and former contract workers" in the thousands who moderated content on the platform and developed post-traumatic stress disorder from the nature of the work.

A 2024 study in European Psychiatry documented secondary trauma among content moderators, detailing the long-term psychological distress caused by years of viewing and suppressing violent or disturbing content.

Content moderation is frequently outsourced by American technology firms, and the workers who spoke to the Guardian were largely based in rural India. 

Monsumi Murmu, a content moderator currently training AI systems, spoke with the paper about how quickly and thoroughly the work affected her, describing sleepless nights and unsettling imagery that looped in her mind's eye at night.

"The first few months, I couldn't sleep. I would close my eyes and still see the screen loading," Murmu recalled, before detailing the distressing themes that haunted her. 

Why is this concerning?

Today, the psychological impacts of content moderation are extensively documented.

Global firm Innodata cataloged the known adverse effects of the work, citing depression, PTSD, anxiety disorders, panic attacks, and adoption of "fringe views" as common outcomes. 

What's the most you'd pay per month to put solar panels on your roof if there was no down payment? $200 or more 💰 $100 💸 $30 💵 I'd only do it if someone else paid for it 😎 Click your choice to see results and speak your mind.

Because content moderation necessarily prevents harmful content from reaching end users, the work is invisible by design, yet workers absorb immense trauma to protect social media users.

AI training — the process of compiling the base of knowledge from which AI tools draw — has been controversial from the outset for several reasons. Initially, the use of copyrighted materials led to lawsuits filed by creators whose work was being used without compensation.

On top of that, AI training is resource-intensive, and big firms such as OpenAI have been loath to disclose how much water and power are used in the process. 

That reticence made far more sense in 2025, after AI data center energy demand caused electric bills to spike nationwide. In the summer, the Department of Energy warned that data center demand was straining an overworked grid, increasing the risk of blackouts.

At the same time, data center pushback escalated, and communities objected to what amounted to ratepayers subsidizing the so-called AI boom.

The Guardian's report told a similar story — technological innovation's costs, be they monetary or psychological, are being passed on to everyday people.

According to the Guardian, only two of the eight content moderators they spoke with received "psychological support" from their contract employers.

What's being done about it?

By late last year, communities across the country had halted $98 billion in new data center construction, showing that local action is an efficient way to take on tech titans.

Get TCD's free newsletters for easy tips to save more, waste less, and make smarter choices — and earn up to $5,000 toward clean upgrades in TCD's exclusive Rewards Club.