Artificial intelligence is already embedded in how your workplace operates. It’s influencing hiring, monitoring performance, allocating work, flagging risk, and supporting decisions that used to sit entirely with people. And for most organisations, the WHS conversation about that hasn’t caught up yet.
That gap is now a legal and operational exposure.
New South Wales has already moved. In addition to existing safe work responsibilities in digitally mediated workplaces, The Work Health and Safety Amendment (Digital Work Systems) Act 2026 formally establishes that digital systems – algorithms, AI, automation, and online platforms – are part of the work environment. Employers now carry a duty to ensure these systems do not put worker health and safety at risk. That includes excessive workloads, unreasonable monitoring or surveillance, and discriminatory decision-making in AI-mediated systems. The question is no longer whether AI creates WHS risk. The question is whether your organisation has looked.
The risk most organisations are missing is digital — and it’s growing.
When AI systems provide fast, confident outputs, people naturally trust them. When dashboards replace conversations, decisions become data-rich but context-poor. When algorithms carry the appearance of objectivity, they gain an authority that often goes unquestioned. This is AI automation bias — and it sits at the intersection of human behaviour and digital systems, which means it won’t show up in traditional risk assessments unless you’re specifically looking for it.
Layered on top of that is cognitive load. Employees navigating multiple platforms, alerts, dashboards, and automated prompts in real time are operating under conditions of sustained strain. The research is clear: a fatigued mind is more likely to default to system outputs. Efficiency, in other words, can silently erode the very human judgement it was meant to support.
The “Three P” priorities frame the response.
The first is to protect the human — recognising that digital systems are now part of the work environment itself, and that the harms they can introduce (from workload intensification to digitally enabled abuse and misconduct) require the same risk management rigour we apply to physical hazards.
The second is to preserve the mind — treating cognitive wellbeing not as a wellness initiative, but as a governance question. Which roles carry the highest decision density? Where is cognitive strain most likely to drive automation bias? These belong in your risk assessments and your procurement decisions.
The third is to power human connection — because relationships are a safety system. They reveal weak signals before metrics do. They create the conditions for people to speak up, challenge outputs, and contribute the context that no algorithm can fully capture.
This isn’t about slowing AI adoption. It’s about leading it well.
The organisations that will get this right are the ones treating safety, culture, and digital transformation as one conversation — not three separate workstreams. That requires leaders who understand both the human and the technical dimensions of risk, and who are willing to move before policy forces them to.
To learn more, download the full whitepaper here.
——-
Yasmin London is a keynote speaker, facilitator, and consultant, and the CEO of First Movers Co — an education and training consultancy dedicated to supporting a human-centred AI era. She works with risk, WHS, safeguarding, and governance leaders to build workplaces that are safer, more resilient, and better equipped for the realities of AI-mediated work. To enquire about speaking or consulting engagements, contact hello@yasminlondon.com
