Philippine Workers Training AI Models Face Psychological Toll from Disturbing Content

Thousands of workers in the Philippines are employed in the hidden labour market that powers artificial intelligence systems, reviewing and labeling disturbing content that includes violence, child abuse, and extremism. Many are now speaking out about the severe psychological toll their work is taking on them. These data labelers and content moderators, often hired through subcontracting firms, work long hours exposed to the most harmful content on the internet so that AI models can learn to recognise and filter it. Workers report symptoms of post-traumatic stress, anxiety, depression, and disturbing dreams. The industry, which pays relatively low wages by Western standards, has concentrated in the Philippines and other developing countries where English proficiency is high and labor costs are lower. Companies often provide minimal psychological support despite the extreme nature of the content workers must review. Campaigners and mental health experts are calling for stronger regulations requiring AI companies to provide adequate mental health support and to compensate workers fairly for the psychological risks they undertake.