The intelligence alliance that cracked Nazi codes in World War II just issued a warning that makes the Enigma machine look like a Speak & Spell. Five Eyes cyber chiefs — representing the United States, United Kingdom, Australia, Canada, and New Zealand — told the Financial Times that AI-powered cyberattacks capable of crippling Western infrastructure aren't some far-off sci-fi scenario. They're months away.
"The timeline is not years, it is months."
That quote came directly from Five Eyes officials, and it deserves to sit there by itself for a second. Not decades. Not "sometime in the future." Months. The same intelligence network formalized under the 1946 UKUSA Agreement — an alliance born from the urgency of wartime code-breaking cooperation — is now telling us that the next generation of cyber weapons will make current hacking look like someone guessing your Gmail password.
The joint assessment laid it out plainly. "Frontier AI models are anticipated to exceed current industry expectations, fundamentally transforming both offensive and defensive cyber capabilities," the cyber chiefs warned. Translation for normal people: the same AI that helps you write emails and generate weird pictures of cats is being weaponized by hostile nations to find holes in power grids, water systems, and financial networks that human hackers would need years to discover.
And we already have proof it's happening. In May, the Google Threat Intelligence Group disclosed that it stopped an attack where AI was used to discover a previously unknown software vulnerability — the kind of flaw that, once found, can be exploited before anyone even knows it exists. The suspected culprits were Russian-linked actors, though Google noted similar activity tied to Chinese and North Korean hacking groups.
Three countries. Three of the most aggressive cyber adversaries on the planet. All racing to turn artificial intelligence into an offensive weapon.
The Trump administration, to its credit, isn't sitting around waiting for the power grid to flicker. The U.S. recently ordered Anthropic, one of the leading AI companies, to restrict foreign national access to its most advanced AI models, citing national security concerns. It's a blunt instrument, but when your adversaries are building digital crowbars, you don't leave the toolbox unlocked on the front porch.
The broader contest here is the U.S.-China AI dominance race, and the Five Eyes assessment suggests the West still holds a defensive advantage — for now. The key phrase being "for now." China has poured staggering resources into AI development, and the intelligence community's own language — "fundamentally transforming" — signals that the current balance of power in cyberspace is temporary. Every month that passes without hardening our defenses is a month China, Russia, and North Korea spend sharpening theirs.
Critics will point out that restricting AI access could slow innovation or hurt American companies competing globally. Fair enough. But that argument assumes our competitors are playing by the same rules we are. They're not. Russia is probing vulnerabilities with AI-assisted tools right now. China has made no secret of its intent to dominate AI by any means necessary. North Korea funds its missile program with cybercrime. The question isn't whether restricting access has costs — it does. The question is whether the cost of doing nothing is higher.
The Five Eyes alliance has been quietly sharing intelligence for eighty years. When all five nations agree on a threat assessment and go public with it in the Financial Times, they're not speculating. They're preparing the public for something they can already see forming.
An alliance built to stop one existential threat is now warning us about the next one. The difference is that this time, the enemy doesn't need to cross an ocean. It just needs an internet connection.
