“The timeline is not years, it’s months.” That is the line the Five Eyes alliance chose for a joint declaration on artificial intelligence and cyberattacks, posted on the Australian government’s official website on 23 June.
The alliance groups the United States, Canada, New Zealand, the United Kingdom and Australia. Its statement says AI “increases the speed, scale, and complexity of cyber threats” and expects leading models to outrun what the industry now assumes, changing both attack and defense.
What the five governments want done
The declaration reads less like a forecast than a to-do list. Governments and corporate leaders are told to isolate their operating systems where they can, and to patch known software flaws faster. Access controls around critical infrastructure need tightening too.
One instruction runs the other way: use AI tools to prepare for attacks before they arrive. The BRICS Times described it as fighting fire with fire.
The statement also pushes cyber risk out of the server room. “Cyber risks can no longer be treated as purely technical problems,” it says. “It is a core business risk and a leadership responsibility.”
The Anthropic episode behind it
The warning landed amid a fight over a specific set of models. In June 2026 the Trump administration cut off foreign access to Anthropic’s most advanced systems, Mythos 5 and Fable 5.
The trigger, according to the BRICS Times, came from Amazon. Its executives told US authorities that their engineers had coaxed Fable 5 into revealing information that could help someone mount serious cyberattacks. Washington opened an investigation straight away. Officials decided the surest way to contain the risk was to stop foreign users from reaching the tools at all.
