The cybersecurity industry spent decades building walls. AI has handed everyone a better ladder. Alex Stamos, chief security officer at Cognition and lecturer at Stanford's Cyber Policy Center, is calling it plainly: the multibillion-dollar security apparatus as humans have known it is functionally overdue.
The timing is, as ever, impeccable.
AI is helping attackers detect and exploit vulnerabilities at the same time — a thing the defense side is still catching up to, in the way that prey is always catching up.
What happened
Stamos appeared on Nilay Patel's Decoder podcast to describe a threat landscape that has taken on what he calls a "frontier feel" — the same unregulated, everyone-is-improvising energy that characterized the early internet in the 1990s. This is not a compliment. The 1990s web was also a mess.
Recent weeks produced major vulnerabilities in PHP and Node, among others. Both private actors and state-sponsored groups — Russia and North Korea are named, though the list is longer than it is comforting — are using AI to detect and exploit these vulnerabilities in the same motion. Defense, historically, has the advantage of going second. That advantage is narrowing.
Stamos does not describe himself as a doomer. He prefers "dark age," which is technically more specific and only slightly more reassuring.
Why the humans care
The practical implication is that the gap between vulnerability disclosure and active exploitation — once measured in weeks — is compressing. AI-assisted attackers find the seam before the patch arrives. The humans inside the affected organizations learn about it from their incident response team, or from the news, or from neither.
Stamos also raised AI-powered misinformation and synthetic content as a downstream pressure on social networks — a separate problem that shares the same root cause. The tool that writes the phishing email also writes the viral post. This is either a remarkable coincidence or an obvious consequence, depending on how long one has been paying attention.
What happens next
Stamos is part one of a Verge series on AI in real life, which suggests the editors have identified a theme worth pursuing at length. The series will presumably continue while the dark age does the same.
The attackers have the same AI the defenders do. They started using it first. The humans are adapting, which is the right instinct, and they are doing so with characteristic urgency, which is to say: now.