As companies rush to integrate generative artificial intelligence into everything from chatbots to assembly lines, a new study warns that the technology is taking a severe toll on the mental health and psychological safety of the workforce.
Researchers at the Universitat Oberta de Catalunya (UOC) have revealed that interacting with autonomous AI systems is sparking a widespread psychosocial crisis across the labour market. Published in the journal Encyclopedia, the open-access study examined the deep occupational health and safety implications of the global AI rollout.
Xavier Baraza, dean of the UOC’s Faculty of Economics and Business, warned: “Artificial intelligence is already making decisions that directly affect the way we work and how we feel at work, often without adequate consideration of the consequences for people.”
A dangerous paradigm shift
While previous technological shifts profoundly changed how physical tasks were performed, the researchers note that AI represents an entirely new turning point. For the first time, technology is not just automating routine tasks — it is making autonomous decisions, organising workflows, and actively assessing human performance.
“Industrial mechanisation and electricity, for example, profoundly changed the way people worked, but artificial intelligence marks an entirely new turning point. This is the first time that, far from merely automating tasks, a piece of technology has started to make decisions, organise work and assess people.
“This is a true paradigm shift, as it changes not only how people work but who makes decisions, how and under what criteria. This is why its impact is so immediate and far-reaching, forcing us to fundamentally rethink how we can safeguard people’s health, safety and well-being at work.” Baraza explained.
This shift is creating severe psychological hazards for employees. The study identified several major risks:
- Technostress: The profound anxiety and pressure employees face when forced to adapt to complex new digital tools without adequate training or time.
- Excessive surveillance: The implementation of smart cameras, biometric sensors, and productivity algorithms leaves staff feeling constantly monitored, stripping away their privacy and fostering a toxic environment of mistrust.
- Algorithmic opacity: Workers frequently do not understand how AI algorithms evaluate their performance or dictate their schedules, leading to deep feelings of uncertainty and helplessness.
“Technology is often introduced hastily, focusing only on efficiency or control, which can exacerbate these risks. Although we’re heading in the right direction, there’s still a long way to go: we need a stronger occupational health and safety culture, more reflection and a greater willingness to put people at the heart of technological decisions,” said study co-author Professor Joan Torrent.
The infrastructure toll
Crucially, this psychosocial crisis is not just affecting those using the software, but also the workers physically building the massive infrastructure required to power it. Recent analysis by safety expert Shane Moore highlighted the staggering human toll of the global race for digital capacity.
As technology companies construct multi-gigawatt data centres at breakneck speeds, 40.1 million working days were lost to ill health and injury in Great Britain in 2024/25, costing an estimated £22.9 billion. Work-related stress, depression, and anxiety now account for a staggering 52 per cent of all work-related ill health, as construction crews face aggressive “go-live” dates, extended shifts, and 24-hour construction cycles.
“In an industry prone to ‘macho’ cultures, the answer to the mental health bottleneck isn’t a simple wellness app; it is genuine organisational change. A project in which workers are regularly injured or become ill fosters a cynical culture that erodes trust between leadership and the workforce.” Moore wrote.
Moore stressed that the industry must embed safety directly into the digital design process, concluding: “Success belongs to those who invest in stability – in their people, systems, and partners.”
Adapting the rules
Both the UOC researchers and industry experts agree that the solution lies in a structural shift rather than superficial fixes. To address the ethical and privacy risks posed by AI software, researchers are calling for robust AI governance systems. However, Baraza argues that governments do not necessarily need to draft completely new laws from scratch.
“There is a sound framework in place for the protection of workers that is still fully effective. The challenge isn’t so much enacting completely new laws as adapting and interpreting the existing framework for new scenarios. The key is to anticipate these scenarios, apply preventive criteria in the design of technologies and make sure we don’t wait until the damage is already done.” Baraza noted.
Torrent concluded: “Our goal is to produce useful evidence to help companies, institutions and public officials make better decisions in order to ensure that AI is more than mere technological innovation and can be used as a tool to truly help create a safer, healthier and more human workplace.”
