Australia's former top cybersecurity adviser has issued a dire warning, claiming the nation remains "dangerously exposed" to hackers following Anthropic's decision to withhold its latest AI model from local authorities.
The model, dubbed Mythos, reportedly possesses an unprecedented ability to identify and exploit vulnerabilities across every major operating system and web browser.
While Anthropic has launched "Project Glasswing"—an exclusive group of US-based tech giants and financial institutions tasked with patching flaws—Australian organisations remain sidelined.
Alastair MacGibbon, who advised the Turnbull government, argued that Mythos has "crossed the Rubicon", rendering traditional security patching obsolete.
He stressed that without immediate access, Australia’s power grids, hospitals, and critical infrastructure are left in a global queue, dependent on Washington’s timeline for protection.
The Albanese government has maintained a cautious stance, recently signing a memorandum of understanding with the American AI firm.
However, Cybersecurity Minister Tony Burke dismissed MacGibbon’s calls for urgent intervention, stating the government relies on current security briefings rather than external commentary.
The domestic friction contrasts sharply with international responses; US and UK regulators have already mobilised high-level summits to defend their banking sectors against Mythos’s capabilities.
Industry experts suggest US legislation may be restricting Anthropic’s ability to share the tool with allies.
With the Commonwealth Bank also awaiting access, the delay raises concerns that while the US secures its own borders, Australian digital defences remain vulnerable to the very exploits Mythos was designed to neutralise.