The risks of using AI and digital technologies at work can be managed using the same risk management process used for any other hazard and risk.
In consultation with workers and their representatives, as well as other duty holders, PCBUs must:
- identify the hazards
- assess the risks (if not already known)
- control the risks
- monitor and review control measures.
Hazards from AI and digital technologies
AI and digital technologies can introduce or contribute to physical and psychosocial hazards
Below is a list of some common examples. The list is not exhaustive and there may be others. Workers may also be exposed to a combination of hazards which may cause harm.
Risks may arise where:
- tasks require repeated and repetitive movements.
- work becomes increasingly sedentary.
- technological devices cause excessive vibration (e.g. from wearable devices).
Risks may arise where:
- AI or digital technologies control the way work is done and introduce a risk (e.g. algorithms pressuring workers to complete jobs faster), or
- where technology physically interacts with workers and directly injures them (e.g. automated security gates malfunctioning or a cobot moving unexpectedly and colliding with a worker).
Risks may arise where:
- systems that set or track the pace of work increase demands, particularly if they don’t provide enough time for the tasks involved.
- where poorly designed or out of date technologies can increase workloads.
- where routine tasks are automated, workers can be left with more complex or cognitively demanding tasks – fewer workers may be needed but their workload may be more intense.
Risks may arise where:
- complex tasks are automated and work becomes less stimulating.
- more tasks are focused on reviewing the output of AI or digital technologies.
Risks may arise where:
- the pace of work is set by computers.
- workers cannot adjust work processes designed to increase efficiency.
- systems closely monitor or set work tasks, particularly if there is little or no input from workers (e.g. if breaks are closely monitored and checked for adherence or if work tasks are set to a strict schedule and monitored).
Risks may arise where:
- technology systems reduce access to supervisors or where technology systems are unintuitive or not fit for purpose.
- systems are poorly designed and make work difficult or confusing to progress.
- workers are not given sufficient training to learn technologies intended to support their work.
Risks may arise where:
- the pace of work and frequency of changes increase because of technology.
- workers are unclear on how to use and interact with technologies for their role.
Risks may arise where:
- AI or digital technologies are not well managed – for example if PCBUs don’t support and train workers in using new technologies, or there is confusion in integrating new systems with old systems.
- The introduction of AI or digital technologies is done poorly or with little consideration or consultation with workers.
Risks may arise where:
- PCBUs use surveillance technologies in an unfair or unjust way.
- automated decision-making processes result in unfair, biased, faulty, or discriminatory decisions.
Risks may arise where:
- online interactions with customers result in higher levels of negative feedback, unrelated to worker performance (i.e. feedback on how digital platforms function – which workers do not control)
- there is a reduction of meaningful human feedback.
Risks may arise where:
- workers work from home or other locations away from others.
- there are fewer social interactions with colleagues or exclusion from team events like team lunches or incidental conversations.
Risks may arise where:
- customer or client abuse increases (e.g. if the technology does not meet user expectations, or if customers or clients are not informed that the technology is being used).
- online and technology-facilitated abuse or sexual harassment occurs (e.g. workers receiving anonymous abuse online or generative artificial intelligence being used to generate abusive or sexually harassing material).
- there is an inability to disconnect when being bullied at work.
- workers are asked to participate in social media marketing and receive online harassment or lead to in-person stalking (e.g. viral social media trends).
Risks may arise where:
- the ability to work from anywhere results in workers being constantly connected to work and working long hours or at times that interrupt sleep.
- work is intensified where workers complete more complex tasks as routine or mundane tasks are automated.
- work is allocated by algorithms without human oversight or consideration of reasonable breaks and workloads.
- technologies are relied on to assess fatigue risks without proper human interpretation or oversight (e.g. only using biomathematical modelling).
Control the risks
PCBUs must eliminate risks, or if that is not reasonably practicable, minimise them so far as is reasonably practicable. Decisions about how to control risks must be informed by consultation with workers. Workers often have a good idea of what would control the risks for them.
For example, to control risks a PCBU might consider:
- choosing software and tools that prioritise safety (e.g., using rostering software which notifies when work is likely to induce fatigue).
- when automating systems or processes, ensure the design avoids work intensification, for example, by allowing workers to alter the pace of work, change tasks or pause work to take breaks as needed.
- engaging IT and WHS expertise to advise on solutions that prioritise safety.
- using digital technologies which are designed to have settings/ options which prioritise health and safety (e.g., adjustable moderation tools where technology is used to interact with others or using technology with accessibility options such as low contrast or colour-blind modes to reduce eye strain).
- Implementing human oversight and audits to ensure technology is working as planned.
- Developing policies and procedures for the use of the digital technology at work, supported by training and supervision.
Maintain and review
PCBUs must maintain and review control measures to ensure they are working effectively. If a control measure is not working effectively, PCBUs must modify or replace it. Reviewing control measures should be done regularly and is required:
- when the control measure is not eliminating or minimising the risks so far as is reasonably practicable
- before a change (such as a software update) that is likely to have new or different WHS risks that the control measure may not effectively control
- if a new hazard or risk is identified
- if the results of consultation indicate a review is necessary, or
- if an HSR requests a review because they reasonably believe one of the above has occurred and it has not been adequately reviewed already.
Other duties
The model WHS Regulations have specific duties that apply to certain hazards, for example, manual handling and psychosocial hazards that apply regardless of whether these risks emerge from the use of AI and digital technologies. Safe Work Australia’s model Codes of Practice on How to manage work health and safety risks and Managing psychosocial hazards at work provide information for PCBUs on ways to control physical and psychosocial risks such as those associated with AI and digital technologies.
Designers of AI and digital technologies also have duties under the model WHS laws to ensure, so far as reasonably practicable, that products are designed with risks to health and safety.
