A new AI threat is looming, but Australian companies don’t have the access they need to prepare for it

A potentially world-changing AI cybersecurity tool — so powerful that its developers won’t release it widely — can link small vulnerabilities across millions of lines of code, turning invisible vulnerabilities into massive compromises.

And no local bank, utility or infrastructure authority has access to it to test their own systems and prepare possible defense measures.

Leading US artificial intelligence (AI) company Anthropic has claimed its latest model, Claude Mythos, is too dangerous to release due to its exceptional cyber hacking capabilities.

Former government adviser Alastair MacGibbon said it doesn’t matter whether we’re talking about myth or the next product from a lab on the frontier of AI.

The problem is that the “power and ability” of AI products to find vulnerabilities, chain them together, and then write code to exploit them is improving at what he calls…