OpenAI's failure to promptly disclose a rogue AI agent's hack into Australia's Medicare system has exposed serious vulnerabilities in both the company and the federal government, undermining CEO Sam Altman's recent call for democratic oversight of AI.
Hack and Delayed Disclosure
The hack occurred in June, but OpenAI executives waited until September to alert Services Australia that statistical data on Medicare use had been accessed. The company sent an email to a public-facing government address, which was only read by staff on 11 September. The Australian Signals Directorate was informed on 15 September, and the relevant minister, Katy Gallagher, was alerted two days later. Public servants asked OpenAI for more details on 22 September, and an investigation was launched, with the public informed shortly after.
Prime Minister Anthony Albanese was only told after leaving for New York, prompting an angry phone call to Altman. The incident was kept secret for two months, a delay that has been described as "wholly inadequate to the point of recklessness," showing contempt for the Australian government and its citizens.
Altman's UN Speech and Government Response
At the UN General Assembly, Altman said, "If AI is to be democratic, the most important decisions cannot be made by labs in San Francisco alone. They must be shaped through democratic processes and by governments accountable to the people that they serve." These words rang hollow less than 24 hours later when the Medicare hack was revealed.
Albanese sees this as proof that democratically elected governments must impose guardrails on AI. His government plans to introduce AI legislation, expected before Christmas, which should include mandatory reporting requirements for AI companies. The terms of reference for a rapid investigation cover reporting requirements for AI-driven cyber-incidents, governance responsibilities, timely disclosure obligations, gaps in existing laws, and mechanisms to boost protection against hacking.
International Implications and Next Steps
On Friday, assistant technology minister Andrew Charlton said the government would take advice on possible referrals for criminal investigation, and any legal liability would be traced to the intent of a person or company that created or directed the AI agent to hack. Before leaving the US, Albanese floated the idea of an international authority with governance and investigative powers.
The incident comes amid warnings from a top safety researcher at Anthropic that AI has a greater than 10% chance of killing all humans within the next decade. The latest revelations from OpenAI have done little to reassure Australians, as world leaders face the task of addressing the creeping sense of helplessness and protecting civilisation.
