25 Years Later

Twenty-Five Years Later

Twenty-five years ago today, al-Qaeda terrorists murdered nearly 3,000 people in New York, Virginia, and Pennsylvania.

For readers who want to relive the day as it unfolded, the 25 Years Ago Live account is reconstructing it minute by minute:

Follow 25 Years Ago Live's minute-by-minute reconstruction of September 11, 2001.

— 25 Years Ago Live (@25YearsAgoLive) September 7, 2026

Our focus here is the present, and a lesson from the years that followed: a threat can be real and still be magnified beyond its actual strength. When that happens, our response can harm us more than our adversary could have managed on its own.

That lesson matters in the emerging campaign to slow or stop AI.

Before we continue, a brief reminder: we’ve refined a successful, defined-risk approach to trading names across the AI stack.

Now back to our post. 

A Small Enemy And A Huge Shock

Before 9/11, Al Qaeda had bombed the U.S. embassies in Kenya and Tanzania and attacked the USS Cole. The 9/11 plot itself was deliberate, disciplined, and years in the making.

It was also the work of a relatively small non-state organization. The 9/11 Commission estimated that the plot cost between $400,000 and $500,000. Nineteen hijackers armed mainly with knives and box cutters turned civilian aircraft into weapons against a country with a military budget then measured in the hundreds of billions of dollars.

Their strategic success came from exploiting a specific assumption. Before 9/11, airline crews were trained to accommodate hijackers, get the aircraft safely onto the ground, and let law enforcement handle the incident. Hijackers traditionally wanted bargaining chips, so passengers had reason to expect that cooperation improved their odds of survival. The 9/11 Commission found that the prevailing doctrine hadn't seriously accounted for suicide hijackings.

That assumption began dying before the attacks were over. Passengers on Flight 93 learned through phone calls what had happened in New York, voted to rush the cockpit, and forced the hijackers to crash in Pennsylvania instead of reaching Washington. They did more to protect America than the fighter planes that were scrambled that day.

Once passengers understood the new objective, the same tactic became extraordinarily difficult to repeat. Hardened cockpit doors added the mechanical answer. The federal government authorized airlines to begin strengthening those doors in October 2001.

This was the central asymmetry of 9/11. The value of al-Qaeda’s vulnerability collapsed once passengers understood the hijackers’ objective and cockpit doors were reinforced. America responded as if it faced a peer with an endlessly renewable strategic advantage.

The Failures Were Specific

The political culture surrounding aviation security also mattered.

During the 2000 presidential campaign, George W. Bush opposed racial profiling. In the second presidential debate, he spoke specifically about Arab-Americans being subjected to “secret evidence” proceedings and promised to end that practice. The documented pledge concerned the use of secret evidence and racial profiling across society. Steve Sailer has argued that Bush’s campaign rhetoric also belonged to a broader political effort against profiling in airport security.

New on SteveSailer.Net — The 9/11 Story Only I Tell. In an October 11, 2000 debate with Al Gore, George W. Bush demanded less airport security scrutiny of Arab passengers. Bush got what he wanted.

— Steve Sailer (@Steve_Sailer) September 11, 2026

Bush’s Transportation Secretary, Norman Mineta, brought a deeply personal history to the subject. Mineta had been incarcerated with his family at the Heart Mountain camp as a ten-year-old during World War II. He later recalled that Bush explicitly said he didn't want Arab-Americans to suffer what Mineta’s family had suffered in 1942. Mineta’s oral history makes clear that he viewed post-9/11 aviation security through the lens of both safety and civil rights. He publicly rejected ethnic profiling in passenger screening.

Then there was Michael Tuohey, the ticket agent in Portland, Maine, who checked in Mohamed Atta and Abdulaziz al-Omari on the morning of September 11. Tuohey later recalled that Atta’s appearance and manner triggered the thought that he looked like an Arab terrorist. Tuohey immediately rebuked himself for stereotyping.

Tuohey’s unease was nevertheless accurate. Gavin de Becker’s The Gift of Fear is built around the opposite habit: notice danger signals and act on intuition before the conscious mind rationalizes them away.

The larger failure was systemic. Atta was selected by the computer-assisted passenger prescreening system, or CAPPS, that morning. Under the rules then in force, the selection only ensured that his checked bag stayed off the aircraft until he boarded. It didn't produce additional screening of Atta himself.

The hijackers also passed repeatedly through the visa, immigration, and customs systems. The 9/11 Commission staff’s terrorist-travel study found that the 19 men had 25 contacts with U.S. consular officers and 43 contacts with immigration and customs authorities. They entered the country 33 times. The study concluded that as many as 15 of the 19 were potentially vulnerable to interception by border authorities.

In other words, America had plenty of contact with the men who attacked it. It lacked the habits and systems needed to turn those contacts into protection. Better visa enforcement, functioning watchlists, intelligence sharing, attention to suspicious behavior, and a locked cockpit addressed the actual vulnerabilities far more directly than occupying countries thousands of miles away.

The Immigration Response Ran The Other Way

The obvious risk-management response after foreign nationals from an Islamist network attacked the United States would have included restricting Muslim immigration. Instead, America expanded it.

Pew Research Center estimated that Muslims grew from about 5% of new green-card recipients in 1992 to about 10% in 2012, or roughly 100,000 legal immigrants that year. Among foreign-born Muslim adults surveyed in 2011, 40% had arrived since 2000.

The demographic result is clearer. Pew estimates that the U.S. Muslim population grew from about 2.4 million in 2007 to 5.5 million today. The same study notes that the United States had no Muslim members of Congress in 2001 and has four today, with a fifth about to join.

New York City now has its first Muslim mayor, Zohran Mamdani, who declared at his inauguration that he was elected as a democratic socialist and would govern as one. A political movement that occupied the margins of American politics in 2001 now controls City Hall in the country’s largest city.

Immigration changes who can build winning coalitions and which ideas can reach power. Mamdani’s faith and ideology are distinct facts; his career reflects the political visibility of a population that was much smaller in 2001.

A Network Became “World War IV”

The political response post 9/11 immediately inflated the scale of the enemy.

Norman Podhoretz eventually called the conflict “World War IV” (the Cold War was World War III in his idiosyncratic classification). Thomas Friedman argued that invading Iraq would send an unmistakable message to the Arab and Muslim world. Television repeatedly showed footage of masked trainees traversing obstacle courses and swinging from monkey bars, while an American general ludicrously compared these spartan camps to the U.S. Marine Corps’ sprawling base at Quantico, Virginia.

The Hamburg cell is a useful corrective. Atta and several central conspirators radicalized and organized in Germany before traveling to Afghanistan. We spent over $800 billion bombing and rebuilding Afghanistan when Al Qaeda in Afghanistan’s only means of attacking us was via immigration. It would have been cheaper to just not let migrants prone to radicalization in.

The same threat inflation appeared after later attacks. Commentators repeatedly searched for an overseas “mastermind” when an attacker used a truck, a knife, or a gun and claimed allegiance to ISIS or al-Qaeda. Many of those links were ideological. In 2022, the director of the National Counterterrorism Center told Congress that 37 of the 45 ISIS- or al-Qaeda-linked attacks in the United States since 9/11 had been inspired by those groups. Those 37 lacked central direction.

That distinction identifies the enemy’s capability. A propaganda brand that inspires an alienated man to rent a truck poses a different strategic problem from a state or military organization able to plan and command attacks at will.

The Mission Became Its Own Industry

A short campaign to punish the Taliban for sheltering al-Qaeda and to destroy the group’s sanctuary made sense. The Taliban regime collapsed quickly, although Osama bin Laden and other senior al-Qaeda leaders escaped. Kabul fell in November 2001; by early December, every major Afghan city had fallen to the Northern Alliance and its U.S. partners. Then footage of starving Afghan refugees—and perhaps motivations similar to those that moved the ticket agent Tuohey—prompted the U.S. and NATO to start a doomed, 20-year nation-building campaign in Afghanistan.

Contractors, consultants, nonprofits, and government offices acquired budgets and careers tied to continuing the mission. Their motives varied; their institutional incentives pointed toward continuation. The resulting constituency outlived the strategic clarity that had justified the first phase of the war. SIGAR’s lessons-learned review describes shifting strategies, unrealistic timelines, weak measurement, and the imposition of Western technocratic models that poorly fit Afghan institutions.

The broader post-9/11 wars produced costs on another scale. Brown University’s Costs of War project estimated in 2021 that the United States had spent or incurred obligations of roughly $8 trillion across the post-9/11 war zones, while about 900,000 people had died directly from the wars’ violence. Those are estimates across several conflicts and countries. They convey the magnitude of the response to an adversary whose greatest attack depended on a vulnerability that passengers began closing within hours.

The AI Parallel

The comparison here concerns policymaking under fear. Al-Qaeda was an organization with an established record of murder. Artificial intelligence is a technology with enormous promise, demonstrated harms, and disputed forecasts of catastrophe.

Yesterday, we examined the campaign to slow AI: a viral resignation warning that labs are gambling with humanity’s survival, a proposed moratorium on new AI data centers, a planned Senate forum, and a well-funded network of AI-safety organizations ready to amplify the message.

The warning could be sincere and the institutional campaign could be real at the same time. The relevant question is how much power the warning should transfer to the people and organizations offering to save us.

The post-9/11 record suggests a demanding test. Name the capability that creates the danger. Identify the vulnerability it exploits. Choose a measure that directly blocks that path. Define the evidence that would show whether the measure worked. Build in an end point.

Flight 93 changed passenger behavior. Reinforced cockpit doors blocked access. Better watchlists, visa review, and intelligence sharing addressed known failure points. Invading Iraq and spending twenty years trying to remake Afghanistan answered a much larger set of ambitions.

AI policy presents the same choice of scale. Security around model weights, testing for specific dangerous capabilities, incident reporting, and rules attached to demonstrable risks are bounded interventions. A data-center moratorium that lasts until Congress constructs a comprehensive regime governing AI products, employment, wealth distribution, local approval, subsidies, labor, and the environment is a political program on a different scale.

Catastrophic forecasts naturally favor the broadest response: if extinction is possible, almost any intervention can be sold as prudent. That logic also makes the new apparatus difficult to limit, because uncertainty becomes evidence for keeping it in place.

The durable lesson of 9/11 is calibration. A small enemy achieved extraordinary leverage when the United States answered a specific security failure with a generational project. AI risks deserve measures designed with evidence, mechanisms, and limits in mind. Fear is a poor substitute for any of the three.