Australia’s national security agencies and critical infrastructure defenders are being granted access to Anthropic’s Claude Mythos Preview AI model, which the company has not released publicly due to its advanced cybersecurity capabilities.

Anthropic announced on Wednesday (AEST) that it would expand access of Mythos Preview to organisations in 15 new countries under its so-called Project Glasswing.

A spokesperson for the Australian Signals Directorate (ASD) told Information Age the intelligence agency welcomed Anthropic’s expansion of the project to include the Australian government and Australian private companies.

An Anthropic spokesperson confirmed Australian organisations would gain access to Mythos Preview, but did not name any specific organisations or disclose the other new countries added to Project Glasswing.

Anthropic had previously only provided the AI model to the governments of the United States and United Kingdom, as well as around 50 mainly American companies and organisations.

That tally will now reach around 200 organisations globally, the company said.

The new additions reflected “several industries that weren’t well represented in our initial cohort, such as power, water, healthcare, communications, and hardware”, Anthropic said.

“And many of the new partners are vendors—companies or nonprofits that maintain codebases that are relied upon by lots of other organisations around the world, including governments.”

Other countries which will gain access to Mythos Preview reportedly include France, Germany, Italy, Switzerland, the Netherlands, Spain, Belgium, Sweden, India, Japan, and South Korea, according to the Financial Times.

The expansion of Project Glasswing will also reportedly see the NATO military alliance, the European Union’s cybersecurity agency ENISA, and some financial market operators gain access to Mythos Preview.

Anthropic publicly announced Mythos Preview in April after the model supposedly found thousands of “high-severity vulnerabilities”, including some in every major operating system and web browser.

Having shared the technology with major companies such as Apple, Google, Microsoft, AWS, and Nvidia, Anthropic said Mythos Preview had since found “more than 10,000 high- or critical-severity security flaws”.

Australian cybersecurity officials and financial regulators warned local organisations to bolster their security posture in the weeks after Mythos Preview’s announcement, when Anthropic reportedly briefed government agencies without granting access to the model.


The Australian Signals Directorate (ASD) says it welcomes Anthropic providing access to Claude Mythos Preview. Image: Australian Signals Directorate

Trump signs executive order on AI testing

Security concerns raised by Mythos Preview and advanced AI models from other firms such as OpenAI contributed to US President Donald Trump's decision to sign a new executive order on Tuesday.

The order asks American firms to voluntary share their most powerful AI models for government cybersecurity testing before they are released publicly.

It directs government agencies to reach agreements with AI developers to test their models up to 30 days before any wider release – a shorter period than the 90-day period reportedly listed in an order which Trump shelved in late May.

The president postponed plans to sign the previous order over concerns that the measure could harm America’s progress in AI.

“We’re leading China, we’re leading everybody, and I don’t want to do anything that’s going to get in the way of that lead,” Trump told reporters at the time.

Anthropic, which has previously clashed with the Trump administration over government access to its AI models, described the president’s new executive order as “an important step in strengthening America’s leadership in AI”.

“We look forward to collaborating with the White House to support its implementation,” the company said.

OpenAI CEO Sam Altman said on Wednesday that Trump’s new executive order “gets the balance right”.

“The US should lead on AI by continuing to develop the very best models, making sure they're safe, and getting cyber tools into the hands of trusted defenders,” he said.

Anthropic files IPO ahead of OpenAI

News of Anthropic’s expansion of Project Glasswing came a day after the company confidentially filed for an initial public offering (IPO), in a race to beat ChatGPT maker OpenAI to the public market.

While Anthropic did not disclose the details of its IPO, the company’s latest funding round saw it raise $US65 billion ($90.6 billion) at a valuation of around $US965 billion ($1.35 trillion).

After fending off a recent legal challenge from Elon Musk, OpenAI is reportedly preparing to file an IPO of its own, which could value the firm at a similar level to Anthropic.

Musk’s own SpaceX, which acquired xAI earlier this year, detailed its own IPO in May with a target valuation of around $US1.75 trillion ($2.4 trillion).