Australia has been granted access to Anthropic’s Claude Mythos Preview, an advanced artificial intelligence model considered too risky for public release because of its powerful cybersecurity capabilities. The system can identify software and network vulnerabilities that could pose serious risks if misused. Access has been expanded through Anthropic’s “Project Glasswing” to about 150 additional organisations across 15 countries, including the Australian government and private companies, after previously being limited to the US and UK governments and around 50 American companies.
The Australian Signals Directorate welcomed the expansion, saying it included the Australian government and local private companies. However, Home Affairs Minister Tony Burke’s office declined to provide details about Australia’s access. Anthropic said the new partners include organisations in critical sectors such as energy, water, health care and communications, as well as vendors whose code is widely relied upon by governments and businesses.
The article places Australia’s access to Mythos within a wider debate about how quickly governments are responding to AI. Anthropic chief executive Dario Amodei, who visited Canberra earlier this year, said AI could strengthen democracies and warned that democratic nations must retain an advantage over autocracies. In the United States, President Donald Trump has signed an executive order asking AI companies to voluntarily submit powerful models for government cybersecurity testing before public release, while resisting what he describes as overly burdensome regulation. OpenAI chief executive Sam Altman welcomed the approach, though OpenAI is also facing legal action in Florida over alleged safety failures linked to ChatGPT.
Independent Curtin MP Kate Chaney is central to the article’s Australian policy focus. She has launched an AI discussion paper warning that Labor is moving too slowly to manage AI’s risks and opportunities. Chaney said: “I think the government's approach to AI is pretty hands-off, it seems to be, 'we'll see what happens, it's going to be fine, it's probably not going to change as fast as everybody says it will', and I think that's really concerning.” She added: “If we want to actually make the most of the opportunities, we are going to need to actually have safeguards in place to build the public trust in AI, and I think that requires a much more active approach than the government is taking.”
Chaney’s paper lists 18 policy priorities, including urgent action on AI-generated child sexual abuse material, risks to children from emotionally intimate chatbots, deepfakes and disinformation. She also wants stronger resourcing for the Australian AI Safety Institute and analysis of likely AI-driven job losses. Another key proposal is making the government’s Data Centre Expectations legally binding, particularly around water and energy use.
The article also highlights growing concern about AI-related data centres. Greenpeace has warned that data centres could demand more new electricity over the next 15 years than Australia’s cars or homes. Assistant Technology Minister Andrew Charlton said public trust is vital, while Treasurer Jim Chalmers argued data centre investment is boosting economic growth but must align with national interests and environmental obligations.