Australia gains access to powerful AI amid questions about readiness

Australia has gained access to Anthropic’s Claude Mythos, which has been deemed too dangerous for public release, as the federal government is urged to speed up its response to the risks and opportunities of AI.

Several private companies and the government are among 150 additional organisations across 15 countries granted use of the Claude Mythos Preview AI model, which the US company has not made widely available due to its cybersecurity capabilities.

Australia’s inclusion in Anthropic’s “Project Glasswing” on Wednesday was welcomed by the nation’s cybersecurity agency, but the government has declined to confirm any specifics around its Mythos access. Labor is also facing calls from an independent MP to step up its efforts to keep pace with the rapid AI advances impacting the economy, workplaces and national security.

Mythos is a powerful AI model able to identify software and network flaws that would pose a significant cybersecurity risk in the wrong hands. As a result, Anthropic has previously only released access to a preview of Mythos to the UK and US governments as well as about 50 American companies, in an effort to help these organisations identify and address any cyber weaknesses.

A spokesperson for the Australian Signals Directorate said they welcomed Anthropic’s decision to “expand Project Glasswing to approximately 150 additional entities globally, including the Australian government and other Australian private companies”. While the office of Home Affairs Minister Tony Burke declined to comment on the Anthropic announcement.