On 15 July 2026, the Prime Minister (PM) announced that Australia will adopt mandatory standards for regulating the use of AI. The PM said: ‘This is our time to decide what AI looks like here in Australia.’
The commitment to adopt mandatory standards is potentially a significant development – if it is backed with new regulatory tools and the capacity to enforce obligations on those developing and implementing AI.
Move from voluntary to mandatory standards
Prior to this announcement, as late as December 2025, the Australian Government was committed to voluntary guidance and the reliance on existing laws to regulate the use of artificial intelligence. This has been described as at a light touch/hands off approach, influenced by weak regulatory models being adopted in the USA and UK and following substantial lobbying by industry.
A large part of this most recent announcement by the PM was devoted to making the case for Australia becoming a world leader in AI investment and innovation. The tone of the speech creates a sense of urgency for government action.
The PM made specific commitments to those in creative industries and journalists. Referring to Pope Leo’s Papal Encyclical Magnifica Humanitas, the PM also acknowledged the economic, legal, social, moral, and spiritual challenge that AI presents.
Importantly, in relation to work, the PM emphasised the need to ‘…create a new generation of good, secure jobs for our economy.’
Soft touch self-regulation not enough
However, we know that left to their own accord, organisations and AI developers who sell products to these organisations, may not be vested with the same goal of creating good, secure jobs. In fact, there is already evidence of AI being used to inappropriately surveil workers and to replace workers in ways that not only devastates these workers, but also produces confusion and anger within the community.
The idea of mandatory standards is a good one. The Centre for Future Work (CFW), along with others, has previously called for mandatory standards around the use of AI at work.
AI at work is ‘high risk’
Work is a vital component of people’s lives, society, and the economy. This is why the potentially disruptive impact of AI at work means it should be treated as ‘high risk’ and managed accordingly.
AI has the potential to replace some jobs and reduce the quality of others. Constant surveillance through AI, and the use of biometrics at work, creates psychosocial hazards and risks and invades workers’ privacy. Traditional consultation measures regarding the introduction of technology have been ignored by some employers, or may be too little too late when considering the scale, speed, and impact of AI use at work.
An issue of trust
At the heart of this problem is the issue of trust. Australians do not trust AI developers, and we do not trust that business will use AI in ways that benefit workers and the community.
Whilst the case for AI at work includes that it increases productivity-including in the PM’s speech, there is currently little evidence that this is the case. In fact, evidence shows that organisations have limited capacity to know what AI is doing at work and how it is actually being used.
A standard for how AI is used at work could help to build trust and ensure that negative impacts on workers are avoided.
Five things the PMs standards should include
To be effective, and to build trust in the use of AI at work, the PM’s mandatory standards should include:
- New workplace rights regulating the adoption and use of AI in the workplace.
This could include a framework in the Fair Work Act with the capacity for this framework to be supplemented by standards at the sector, industry, or enterprise level through collective bargaining. Workplace rights should include worker (and union) engagement in decision making and use of AI at work. There should be mechanisms within the Fair Work Commission to address disputes over the introduction and use of AI at work. - The adoption and use of AI has created new workplace hazards and risks.
Thus, there should be a primary duty on organisations to address and mitigate the discriminatory, and health and safety, risks posed by the application of AI. This should be supported by the adoption of new model regulations and codes, and greater rights for health and safety representatives and unions to inspect AI systems at work to ensure they are safe. - The Fair Work Act should be amended to address workers’ privacy concerns related to AI.
The collection of workers’ personal and sensitive information and worker surveillance should only be done as last resort. Organisations should be required to justify their need for surveillance and for the collection of personal and sensitive information from workers. There should be restrictions on how organisations can use workers’ personal and sensitive information and on how, when, and why surveillance technology is used at work. - No workers should have their livelihoods determined by AI decision making.
As an overarching principle, in any standards for AI at work, there should be an obligation of ‘human in command’ decision making and a right for workers not to be subjected to automated decision making where a decision significantly affects their lives. - Clearly there is likely be both job creation and displacement arising from the use of AI at work.
Therefore, a digital just transition, which allows workers, whose jobs may be displaced by the application of AI at work to plan and reskill for new jobs, should be created.
The actions of organisations regarding the implementation of AI at work thus far demonstrate that ‘soft’ law mechanisms and self-regulation should not be the primary form of managing the use of AI at work. Thus, to be effective, any standards on AI at work should be backed by specific legislative provisions.
The goal of regulating AI in the work context should be directed to the protection of workers and a desire to achieve decent work. Promoting AI innovation must not overshadow the need for decent jobs and fairness at work now and in the future.