Philip Greenspun’s Weblog

The 9/11/2001 jihad was 25 years ago. See A visit to the United Flight 93 crash site for last year’s post in remembrance. Today, though, we’ll look at the intersection between jihad and artificial intelligence.

The U.S. response to the explicitly Islamic attack on 9/11 was to declare that Islam was the “Religion of Peace” and that we would, therefore, use our military to try to find the, perhaps, 1 Muslim out of 10,000 that might be somehow involved in violence, contrary to all Islamic principles as we understood them. We would also expand Muslim immigration to the U.S. because, again, only 1 out of 10,000 might want to perpetrate a spectacular attack on Americans and it would be straightforward for our police state to find this tiny minority before anything bad happened.

The latest “silver bullet”, so to speak, in the “sift through what previously would have been called ‘the enemy’ one by one and occasionally arrest or kill someone” is to build a computer program that will sift through what previously would have been called “the enemy” one by one. In other words… Artificial Intelligence.

Project Maven: A Marine Colonel, His Team, and the Dawn of AI Warfare by Katrina Manson, a Financial Times and Bloomberg journalist, covers a U.S. military project that began in 2017 and, despite some initial teething pains, has proved useful against Islamic opponents (Somalia, Afghanistan, Iraq, Syria) and for helping Ukraine find targets in the big conventional conflict of the present (versus Russia). One of the biggest challenges with development was that software experts in Silicon Valley had no allegiance to the United States and, therefore, weren’t interested in helping the U.S. military. “‘The Business of War’: Google Employees Protest Work for the Pentagon” (NYT, April 2018):

Thousands of Google employees, including dozens of senior engineers, have signed a letter protesting the company’s involvement in a Pentagon program that uses artificial intelligence to interpret video imagery and could be used to improve the targeting of drone strikes.

The letter, which is circulating inside Google and has garnered more than 3,100 signatures, reflects a culture clash between Silicon Valley and the federal government that is likely to intensify as cutting-edge artificial intelligence is increasingly employed for military purposes.

Google eventually shut off the Department of Defense from its cloud services and abandoned the DoD as a video analysis customer. The exception to the globalist “we’d be just as happy to help China, Hamas, or anyone else” perspective was Palantir under Alex Karp‘s unabashedly pro-America leadership. The result of Karp being an outlier was that Palantir achieved the dominant contractor role for military AI software (Anduril is today a serious rival, however). After Palantir’s years of work, the Trump v2.0 administration was able to deploy the software to monitor people crossing the southern border, to help find suspicious boats to summarily sink, and to pick targets in our “slow-motion surrender to Iran” project.

The book is worth reading if you want to understand how warfare is being transformed and also just how cumbersome and inefficient the U.S. military is. Even today, despite Pete Hegseth and Donald Trump being all-in on AI, our military budget for AI is tiny compared to what we’re spending on legacy technology and personnel and on what commercial companies are investing in AI, e.g., to build data centers. (For most of the time period covered by the book, the AI budget was about 1% of what we’re spending on veterans, such as monthly checks to 100% disabled Graham Platner).

The author of Project Maven never questions the effectiveness of the “sift through a large civilian population, many of whom likely support the fighters, one by one” approach to warfare. Let’s step back, though, and consider whether this approach, including in its AI-enhanced form, leads to decades-long or centuries-long war. From New Yorker 2010, regarding the founder of modern nursing:

Humanitarianism also had a godmother, as Linda Polman reminds us. She was Florence Nightingale, and she rejected the idea of the Red Cross from the outset. “I think its views most absurd just such as would originate in a little state like Geneva, which can never see war,” she said. Nightingale had served as a nurse in British military hospitals during the Crimean War, where nightmarish conditions—septic, sordid, and brutal—more often than not amounted to a death sentence for wounded soldiers of the Crown. So she was outraged by Dunant’s pitch. How could anyone who sought to reduce human suffering want to make war less costly? By easing the burden on war ministries, Nightingale argued, volunteer efforts could simply make waging war more attractive, and more probable.

Nightingale (1820-1910) seems to have been amply proven correct by the Arabs who call themselves “Palestinians”. They’ve stayed at war for more than 75 years, 100 percent of their basic needs (shelter, food, health care, education for as many children as they want to have, etc.) paid for by humanitarian aid. Yemen seems also to prove Nightingale correct. They’re 12 years into a civil war in which nobody has to concede defeat because Western aid keeps pouring in. No matter how many times subsets of Yemenis threaten or attack Western countries, none of these countries will follow the proven approach of flattening enemy cities and cutting off food shipments to the enemy (even if the Yemenis did not surrender, they’d have to concentrate on agriculture rather than war in order to avoid starvation) because… we have no enemies. We start from the assumption, never questioned, that 99.99% of Yemenis love everything about the West. The attacks on the West will stop when we have used our fancy technology, including AI, to find the 0.01% of Yemenis who aren’t our best friends.

With the rise of precision weapons and targeting policies that forbid significant harm to civilians, there is no reason for a civilian population to choose peace, especially when combined with a policy that anyone at war is entitled to massive shipments of humanitarian aid. Not everyone in a society will be brave enough to carry a rifle, of course, but there won’t be any downside to allowing those who do carry rifles to live in one’s apartment building, set up headquarters in the local hospital, store ammo in nearby schools, etc. The Germans and Japanese repudiated war for generations after they suffered a defeat by non-precision weapons. Civilians who can rely on the enemy’s AI refraining from killing women and children are civilians who can encourage and support their fighters for decades if necessary.

What would we have done after the 9/11 attack if we didn’t take this approach to fighting back and didn’t presume that everyone loves us except for perhaps a tiny percentage of some population? The question doesn’t make sense because, absent the idea that only a negligible percentage of Muslims support jihad, there almost certainly wouldn’t have been a 9/11 attack. After the 1993 World Trade Center jihad, we would have shut down immigration by Muslims, flights to Islamic countries, and temporary visas for most Muslims. Here’s how the 1993 jihad unfolded, according to Wikipedia:

Yousef arrived illegally in the United States on September 1, 1992, traveling with Ahmed Ajaj from Pakistan, though both sat apart on the flight and acted as though they were traveling separately. Ajaj tried to enter using a forged Swedish passport, though it had been altered and thus raised suspicions among Immigration and Naturalization Service (INS) officials at John F. Kennedy International Airport. When officials put Ajaj through secondary inspection, they discovered bomb-making instructions and other materials in his luggage, and arrested him. The name Abu Barra, an alias of Mohammed Jamal Khalifa, appeared in the manuals. Yousef entered on a false Iraqi passport claiming political asylum, and was given a hearing date.

(Imagine a group of people dumb enough to run an immigration/asylum system that allowed a foreign soldier to stay in the U.S. long enough to attack the U.S. and then dumber enough to continue operating that immigration/asylum system even after the attack.)

Our response to both the 1993 and 2001 jihads was to expand the pool of potential jihadis resident in the U.S. (current Muslim population is three times the 1993 level) and legally visiting the U.S. (All of the 9/11 hijackers received visas to enter and stay in the U.S.; a handful of them overstayed their visas, but no effort was made to deport them.) We drew from a pool in which roughly one third of those polled said that the 9/11 attacks on the U.S. were either “somewhat,” “largely,” or “completely” justified (Gallup World Poll of Muslims; a respondent who answered that the 9/11 attacks were “mostly justified” was classified as a “moderate”).

We’ve had multiple domestic jihads already in 2026, all of them perpetrated by U.S. citizens (Emir Balat and Ibrahim Kayumi; Mohamed Bailor Jalloh; Ayman Mohamad Ghazali; Ndiaga Diagne). The book was written before these jihads occurred, but the author cites Aimé Césaire’s “imperial boomerang” idea, in which technology developed to deal with a restive colony will eventually be deployed back home. In other words, Americans desperate to be safe will accede to the use of Maven-like technology domestically, e.g., to ingest all security camera footage and follow suspected attackers. It is unclear that this would have helped with all of the 2026 jihads. Only Mohamed Bailor Jalloh had ever been looked at by the FBI (he was actually imprisoned for eight years and then we were wise enough to release him back into U.S. civilian life rather than stripping him of the citizenship that we were wise enough to give him and deporting him to his native Sierra Leone; certainly there wouldn’t be any reason to update our laws so that people convicted of terrorism automatically had their naturalization reversed). Perhaps the answer will be to track all residents of the U.S. via a Maven-style AI. Cash will be outlawed so that any purchase of any material useful in making a drone or a weapon can be tracked. For drone ingredients, especially, whatever is purchased will need to have some kind of signature to make it possible to track a homebuilt drone weapon back to the buyer/builder. What if someone 3D-prints parts, you might ask? 3D printers will have to be outlawed for consumer ownership and/or include built-in reporting functions.

Monica Rodriguez Smith, killed in the 1993 World Trade Center jihad on what was supposed to be her last day of work before maternity leave:

(The death of her “child” was and is considered the result of a criminal act, yet a pregnant person at seven months of pregnancy today is able to receive abortion care in Massachusetts for any reason at all (one doctor’s judgment) and in New York if a health care provider determines that abortion care would improve the pregnant person’s “mental health”. The human who is subjected to abortion care is, in that circumstance, considered a “fetus” rather than a “child”.)

An MIT graduate school classmate:

According to the 9/11 Commission Report, [Danny] Lewin was stabbed by one of the hijackers, likely Satam al-Suqami, who was seated directly behind him. The Commission speculated that Lewin, being a veteran of the Israeli military, may have attempted to confront Atta or Omari, who had been seated in front of him, unaware that al-Suqami was sitting behind him.

Photos summarizing the change in U.S. society:

9/11/2001, in what Mayor Mamdani calls “Palestine”: