Signals intelligence agencies from Australia, the US, the UK, New Zealand and Canada have issued an urgent warning: powerful AI tools that can launch devastating cyber attacks are only months away. They point out that while AI will eventually bolster cyber defenses, it also speeds up and deepens offensive capabilities. Their joint statement calls for leaders to treat cyber risk as a core business concern, not just a technical problem, and to coordinate responses across entire organisations and societies. Much of the spotlight is on Anthropic’s Fable and its predecessor Mythos—models designed to find and even exploit security flaws. The US government has already barred foreign nationals from using both tools, citing national security. Experts warn that other states or companies, including China, could soon roll out similar or more advanced systems. Meanwhile, Australia has signed a non-binding deal with Anthropic to share AI progress and encourage safety, reflecting a “light-touch” regulatory approach aimed at reaping economic gains.