September 24, 2026 — 5:33pm

OpenAI’s breach of a Medicare website reveals just how unprepared Australia is to defend against the rising risks of autonomous artificial intelligence.

Yesterday, the Herald addressed the importance of regulation following US President Donald Trump’s outright rejection of international regulation for what he has rebranded “super intelligence”.

The Medicare breach, which OpenAI claims was unintended but admits may have infiltrated other Australian government portals in NSW and Victoria, demonstrates how urgent regulation is.

The fact that an autonomous AI agent bypassed security barriers to gain access to non-public data underscores the new risks confronting our national infrastructure.

Prime Minister Anthony Albanese rightly expressed “extreme concern” to OpenAI CEO Sam Altman and established a multi-agency taskforce to investigate the breach, which may be referred to police.

The incident shines a light on a much bigger challenge: government investment in cybersecurity and AI governance has not kept pace with the burgeoning risks posed by rapidly advancing AI, leaving national systems under-resourced.

Tellingly, the Medicare breach occurred in June but went undetected in real time. Federal ministers learned of it this month only after OpenAI self-reported the incident to a general Services Australia inbox.

Alarm bells have been sounding for some time: a raft of tech leaders, researchers and whistleblowers have put on record concerns about the dangers of losing human control over autonomous AI.

Among positive steps, the Australian AI Safety Institute has been established to test and evaluate the risks of AI models, although critics argue the agency is severely underfunded and lacks authority.

The government is also actively encouraging more AI development and infrastructure onshore, aimed at giving Australia greater expertise and influence in regulation.

Voluntary guardrails and self-regulation are clearly not enough. As big tech companies have made clear in their stewardship of social media platforms, regulation is essential to enforce compliance and protect the public interest.

Human oversight and accountability will also be crucial, given autonomous AI agents pursue tasks without an inherent understanding of boundaries.

Australia already faces hostile cyber threats from the likes of Russia, China, Iran and North Korea. ASIO Director-General Mike Burgess warned last year that the nation had entered an “era of strategic surprise and security fragility” with a heightened risk of high-impact sabotage.

That threat landscape has broadened to include breaches originating from allied nations. Even if OpenAI’s model acted without malicious intent, an out-of-control agent from a trusted partner highlights a profound new vulnerability and must serve as a wake-up call.

The risks to our security, safety and privacy are real. Without decisive action and investment in advanced defence systems, Australia will be vulnerable to future incidents that may not be as benign as the Medicare breach.

The taskforce into OpenAI’s breach will put existing processes under the spotlight and consider how to ensure such incidents don’t happen again. The Albanese government in turn must dedicate greater resources while maintaining its firm stance on international regulation.

Start the day with a summary of the day’s most important and interesting stories, analysis and insights. Sign up for our Morning Edition newsletter.