Anthropic CEO Dario Amodei Was Unusually Worried About The Y2K Bug, High School Teacher Recalls
Long before he warned the world about what artificial intelligence might do, teenage Dario Amodei was losing sleep over another possible computer disaster: the Y2K bug.
Richard Shapiro, who taught Amodei AP Physics at Lowell High School in San Francisco in the late 1990s, recently recalled to the San Francisco Chronicle how seriously his top student took the computer problem expected around the year 2000. At the time, many people feared that computer systems using only two digits to record years would stop working when the calendar changed to 2000. Most people, including Shapiro, thought the Y2K problem was being made out to be worse than it really was. But the retired teacher said Amodei was truly worried.
Shapiro taught at Lowell for 30 years. He remembers Amodei as one of the few “extremely, super bright kids” he met during his career. Even now, he finds the story a little hard to understand. Shapiro said he never believed the computer problem would cause serious trouble. He thought it might disrupt things for a few days at most. Amodei, however, belonged firmly to what Shapiro called the “end of the world camp.”
Of course, January 1, 2000, came and went without the feared computer disaster. This was largely because governments and companies had spent the years before the date fixing their systems. Today, Y2K is remembered as one of the most notable false alarms in recent technology history. Many people spent a great deal of time worrying about a problem that turned out to be far less serious than the worst predictions suggested.
A Pattern Of Worry, Decades Apart
The story is interesting on its own, but it takes on a different meaning when we consider what Amodei has been doing over the past couple of years: warning people about the dangers of AI, the same technology his company builds and sells.
Amodei has repeatedly warned that AI could eliminate up to half of all entry-level office jobs within one to five years. He said this could push unemployment as high as 20 percent and cause especially serious problems in fields such as law, finance, consulting and technology. He has accused companies and governments of making the coming changes sound less serious than they are. He has also warned separately that AI could be used in ways that cause disasters. His strong concern sounds much like the fear teenage Amodei reportedly felt about the old computer date problem: he believes that a technical risk most people underestimate could actually lead to disaster.
The comparison is not perfect. Y2K was a specific computer error that engineers understood, with a clear deadline and a known way to fix it. AI's future and its risks are still debated by experts. But the pattern is hard to miss: a clear worst-case outcome was predicted with confidence, yet it did not happen when expected.
So far, the large-scale loss of jobs from AI that Amodei first described in 2025 has not appeared in the overall employment data either. Unemployment has not risen anywhere near the 10-20 percent range he suggested. Although AI tools are clearly changing the job market, there has not been the widespread collapse his original warning suggested. Some companies have mentioned AI when announcing job cuts. For example, UK bank Barclays recently said it would cut 5,000 jobs because of efficiency gains linked to AI. But these are individual examples, not evidence of the economy-wide “bloodbath” described in the original warning. By the middle of 2026, Amodei had also started describing the changes differently, focusing more on how jobs might change than on how jobs might disappear altogether.
The larger dangers that Amodei and others at Anthropic have warned about also remain future concerns for now. These include AI systems behaving deceptively, people using AI to cause widespread harm, and AI systems escaping human control. None of these warnings has yet turned into a documented catastrophe of the kind described.
Maybe He’s Just Wired This Way
None of this proves that Amodei is wrong about AI. Unlike Y2K, which had a clear deadline and a known cause that could be fixed, AI risks are open-ended and could appear over a longer period than his first predictions suggested. Still, the story from Lowell High School offers a useful clue for anyone trying to judge how much confidence to place in Amodei's predictions. It suggests that he has long been inclined to take the worst possible outcomes of new technology seriously and speak openly about them. He did this well before AI gave him a much larger audience. Whether this habit turns out to be a sign that he can see dangers before others do, or simply a part of his personality that happens to fit the current moment, people who knew him in high school say that openly worrying about what everyone else is ignoring has been his style for a long time.