AI regulation takes familiar industrial relations battle lines

The Albanese government is under pressure from the union movement to regulate AI strongly to ‘embed workers’ rights’ with a standalone AI Act.

OpenAI chief executive Sam Altman. Picture: AFP

You can now listen to The Australian's articles. Give us your feedback.

You can now listen to The Australian's articles.

The Albanese government is under pressure from the union movement to regulate artificial intelligence strongly to “embed workers’ rights” with a stand­alone AI Act, a move the Productivity Commission says would “increase the complexity and regulatory burden”.

In a submission to the government’s AI regulation consultation, the ACTU has asked the government to create a new public authority to co-ordinate AI policy across government and for the term “high-risk AI” – inviting the most regulatory oversight – to apply to any AI system that affects “work processes, working relationships, and the rights and conditions or working people”.

AI manufacturers, banks and other large companies, however, have pressed the government for lighter touch regulation.

The Business Council of Australia said the EU and Canada’s approach to AI regulation had resulted in “businesses pulling out due to regulatory uncertainty and their inability to comply with the onerous obligations”.

The Albanese government has previously indicated an intention to pass “mandatory guardrails” to regulate burgeoning AI tech­nology, especially in high-risk ­applications.

This could oversee applications of AI in a range of industries such as self-driving cars, individually targeted advertising, or in helping prepare legal advice.

“We know Australians are concerned about AI in high-risk settings; doing nothing isn’t an option,” Industry Minister Ed Husic said. “Government is taking on board the views from these submissions and will announce next steps soon.”

The ACTU said it most favoured a standalone AI Act “due to the greater level of co-ordination it offers, its emphasis on also covering the responsibilities of developers of AI systems with enforceable regulation, and as it better reflects the urgency required to ensure protections are in place”.

This was supported by the Media Entertainment and Arts Alliance and the Australian Human Rights Commission.

The Productivity Commission and the Australian Competition & Consumer Commission said they opposed a standalone AI Act, spruiking a “framework approach”. This would see the parliament pass legislation that could broadly define mandatory AI guardrails and a consistent set of definitions, which would then be implemented through amendments of separate legislation and subordinate regulation.

The ACCC said a standalone AI Act might not provide “flexibility, may lead to regulatory incoherence, and could increase the burden for businesses”.

Business groups, AI manufacturers and large companies largely opposed a standalone AI Act and favoured the framework approach, or simply amending relevant legislation.

Atlassian, for example, said a framework approach would provide a “unified set of principles and governance structures while still allowing for sector-specific adaptation”.