A Rare Joint Statement From Intelligence Agencies
On June 23, CNN reported, the US and its "Five Eyes" intelligence partners (US, UK, Canada, Australia, New Zealand) issued a joint warning: AI's attack capabilities are strengthening rapidly, and malicious actors may breach the network defenses of governments and large enterprises within "months." Intelligence agencies rarely put it this bluntly or give a timeline this short, and that itself is a signal.
Why "Months"
The basis for the judgment is the asymmetry in how fast attack and defense adopt AI. Attackers can use AI to automatically find vulnerabilities, write exploit code, and launch social engineering in bulk—the barrier and the time required are both dropping; whereas defenders deploying the same capabilities often have to pass through compliance, audit, and staffing gates, and are much slower. Between the one that's fast and the one that's slow, the window left for defenders gets compressed to the order of "months." Looked at alongside another recent thread, the picture is more complete: in the same week, OpenAI had just released GPT-5.5-Cyber, available only to vetting bodies, and Anthropic's powerful model was also brought under government control over safety concerns. Everyone is racing for the high ground of "AI offense and defense"—the capability is the same set, and whether it's used to defend or to attack comes down entirely to who gets it first and how it's gated. For enterprise security teams, the practical meaning of this warning is simple: stop treating AI-assisted defense as a "later" matter.
via: CNN