The Five Eyes intelligence alliance has warned that artificial intelligence models are evolving at a rapid pace and could surpass current cybersecurity capabilities "within months."
The risks associated with AI-assisted hacking came to the fore after American company Anthropic announced in April that two of its advanced models possessed unprecedented capabilities for detecting software vulnerabilities, prompting the White House to intervene and push the company to suspend access to them.
Security agencies from the United Kingdom, the United States, Australia, Canada, and New Zealand have urged governments and companies to move quickly to prepare for the evolution of artificial intelligence.
The Five Eyes alliance countries said in a joint statement that "the pace of developments in advanced AI means that assumptions about cyber risks risk becoming outdated within months rather than years."
They added that artificial intelligence "lowers barriers for malicious actors and increases the speed and complexity of attacks."
The statement continued: "Vulnerabilities will inevitably emerge, but preparedness allows them to be contained quickly and prevented from escalating into major operational and financial crises."
The alliance recommended that organisations in particular integrate AI tools into their security operations, update legacy systems, and restrict access to critical systems.