MACHINE-SPEED TERROR

AI, SEPTEMBER 11, AND THE CIVIC DUTY TO PREPARE BEFORE THE SHOCK

Twenty-five years after September 11, the most important lesson may be larger than aviation security or counterterrorism: a society must think seriously about low-probability, high-consequence risks before catastrophe makes careful debate almost impossible.

That lesson now belongs in the artificial-intelligence conversation.

This week, former Anthropic researcher Jacob Coxon told WIRED that many people building frontier AI believe the next year or two could be decisive for humanity. Coxon pointed to two pathways that deserve special public attention: AI-assisted biological threats and AI-enabled attacks on critical infrastructure. He also warned that competitive pressure among companies and countries can reward speed even when researchers themselves want stronger safeguards.

OpenAI chief scientist Jakub Pachocki delivered a parallel warning from inside another leading laboratory. In "An Alien Mind," he wrote that today's systems can already operate computers, collaborate with people and other AIs, conduct research, and transform computer security. He expects AI to play an increasing role in improving AI itself, and argues that no laboratory has yet solved alignment and monitoring well enough to keep scaling at maximum speed indefinitely. His proposed answer combines better safeguards with shared safety thresholds, auditing, government involvement, and international coordination.

Pachocki also raises a civic problem: concentration of power. If work that once required thousands of experts can eventually be performed by a few people controlling enormous computing systems, then access to advanced AI could concentrate scientific, economic, military, and political capability in unusually small hands. Safety is therefore also a democratic question about who controls these systems and who can challenge their decisions.

At the same time, Treasury Secretary Scott Bessent has framed AI leadership as a national-security competition with China, warning that if China pulled decisively ahead, "nothing else would matter." That language captures the dilemma. The United States has strong reasons to remain technologically capable, yet a race understood as existential can make restraint harder. Coxon and Pachocki are effectively asking whether the world can compete without letting competition erase the brakes.

This is where September 11 offers a civic warning.

It is reasonable to consider a future "9/11-style" event in which AI is not the mastermind but the capability multiplier: helping a person or group find vulnerabilities, automate cyber operations, design dangerous materials, coordinate machines, impersonate trusted voices, or overwhelm defenders with speed and deception. The evidence does not justify saying such an event is inevitable. It does justify preparing for the possibility that AI could make a strategic shock larger, faster, cheaper, or harder to attribute. Coxon himself identifies critical-infrastructure hacking and biological misuse as examples of pathways that worry him.

The second danger begins after the shock.

A frightening AI-enabled attack could create enormous public demand for protection. Government agencies might then turn to the same technology for surveillance, automated threat scoring, identity verification, cyber defense, border screening, emergency information control, or autonomous security systems. In shorthand: AI could help generate the shock, and AI could then be invoked to govern the aftermath.

Some of those tools may save lives. The civic question is whether emergency systems are transparent, lawful, reviewable, temporary where appropriate, and subject to meaningful human judgment. A democracy should not wait until panic to decide what powers machines may exercise over people.

The practical agenda is resilience: stronger cybersecurity, independent AI safety testing, incident disclosure, protection of critical infrastructure, international safety agreements, privacy safeguards, due process, clear human accountability, and public institutions technically capable of supervising the systems they may someday depend upon.

September 11 teaches that preparedness is not fear. It is stewardship. The time to build guardrails is while society still has the freedom to build them carefully.