ACTU calls for AI rules after OpenAI agent hits Medicare

The Australian Council of Trade Unions has called for urgent regulation of frontier artificial intelligence models after an OpenAI agent breached the Medicare system, describing the incident as a turning point for AI oversight in Australia.

The ACTU said the breach involved an AI agent developed by OpenAI and was a serious incident affecting a core public service. It said the Prime Minister had advised that the incident took place on 18 June, that OpenAI became aware of it in August, and that Services Australia was not notified until 10 September.

The breach has intensified concerns about how advanced AI systems are governed and how quickly companies disclose incidents involving public institutions, according to the ACTU. The union movement has argued since 2024 for stronger rules covering frontier AI models, particularly those built by large overseas technology groups.

Regulation push

The ACTU is urging the Australian Government to create a mandatory licensing regime for frontier AI models and give the Artificial Intelligence Safety Institute formal regulatory powers. It said the Medicare incident had strengthened its case for closer oversight of the most advanced AI tools before and after deployment.

The union body said the issue went beyond a cyber incident, raising broader questions of national control, accountability and public trust. It also backed Prime Minister Anthony Albanese's criticism of OpenAI, while arguing that political condemnation alone would not address the risks posed by the technology.

Joseph Mitchell, Assistant Secretary of the ACTU, said the episode had exposed a gap between the pace of AI development and the pace of regulation.

"Unions are outraged to learn this morning that an OpenAI agent has breached Medicare's website. Medicare is a fundamental union accomplishment and vital to the lives and wellbeing of all Australians," Mitchell said.

He linked the incident to broader warnings from organised labour about advanced AI systems.

"Unions have warned for years that we need appropriate regulations to keep Australians safe from the emerging capabilities of these powerful AI models, being developed by large and powerful tech companies in the United States and China," Mitchell said.

Disclosure questions

The timeline cited by the ACTU is likely to sharpen scrutiny of disclosure practices when AI systems are involved in breaches of public infrastructure or government-linked services. The organisation said the gap between the June incident and the September notification to Services Australia showed why regulators needed clear authority and mandatory reporting rules.

The comments also reflect a wider debate in Australia over whether existing digital and cyber rules are sufficient to address autonomous or semi-autonomous AI systems. Policymakers have been weighing how to supervise a fast-moving field largely shaped by companies headquartered outside Australia.

Mitchell said the Medicare breach should not be viewed as an isolated matter.

"This is a turning point for AI regulation. That this is the first disclosure does not mean it is the only instance. The AI companies have shown us that they cannot be trusted to properly manage the risks of their own powerful models," Mitchell said.

The ACTU's preferred approach would place frontier AI models under a licensing framework administered by the Artificial Intelligence Safety Institute. That would shift the institute from an advisory or research role to one with direct authority over access to the Australian market for certain advanced systems.

The proposal reflects a broader international debate over whether the largest AI developers should face pre-release testing obligations, licensing conditions and stricter incident reporting duties. In Australia, the Medicare case may add momentum to calls for a system tailored specifically to advanced general-purpose models rather than relying on existing consumer, privacy and cyber laws alone.

Mitchell restated the union movement's position on the framework it wants the government to adopt.

"Australian unions reiterate our call on the Australian Government to create a licensing regime for powerful frontier AI models to be regulated by the Artificial Intelligence Safety Institute to keep Aussies safe," Mitchell said.

The ACTU also argued that the episode had damaged public confidence in AI at a time when businesses and governments are testing broader uses of generative systems and autonomous tools. Concerns about reliability, misuse and accountability have become more prominent as AI products interact with sensitive data and essential services.

"We need to regulate these companies. We need to protect against the dangers of their products. And we need to make sure they are not undermining the Australian institutions that uphold our quality of life. If you want to know why AI has no social licence in Australia, look no further than this appalling infiltration of Medicare by an AI agent," Mitchell said.