Editor’s Note: This story is one in a three-part series commemorating the 10-year anniversary of Pizzagate, a conspiracy theory that led to a shooting at the Washington, D.C., restaurant Comet Ping Pong in 2016. Read Kaitlyn Tiffany’s analysis on the legacy of Pizzagate and Charlie Warzel’s profile of the Pizzagate shooter, Edgar Maddison Welch.

James Alefantis had an excruciating hangover when he first learned that his restaurant, Comet Ping Pong, was the suspected site of a global sex-trafficking organization. It was late October 2016, and he’d just thrown a 10-year anniversary party for the Washington, D.C., pizzeria. There had been much to celebrate: After grinding through a few slow years, Comet had become a favorite among families in the city’s prosperous Northwest quadrant and also, improbably, an important venue for punk bands. Alefantis had hired a band to play the party and, like many of the hundreds of guests who showed up, he overindulged at the open bar. The next morning, he was barely conscious when his general manager called to say that a reporter had been in touch with some questions about a thing called “Pizzagate.”

Alefantis was, as you might imagine, surprised to hear that people on 4chan and Reddit had identified Comet’s basement—which does not exist—as the headquarters of a pedophilia ring run by Bill and Hillary Clinton. “It sounded like the dumbest thing ever,” he told me earlier this month. The theory was based on a bizarre reading of leaked emails sent by John Podesta, who was the chair of Hillary Clinton’s presidential campaign at the time, and knew Alefantis socially. Cheese pizza was said to be Podesta’s code word for “child pornography,” and pasta his word for “little boy.”

Alefantis’s first instinct was to brush it off. He figured that Pizzagate was just another one of those downs that tend to follow ups in the restaurant business, like when you have a great weekend and then on Monday, your hood fan goes on the fritz. People say lots of crazy things online, especially when the national mood is inflamed by a presidential election. After Donald Trump won a few days later, Alefantis allowed himself to hope that the fever would break.

The decor at Comet can be bewildering for first-time visitors. The dining room is dominated by an enormous, fire-breathing pizza oven. A mint-green Vespa hangs, inexplicably, from the ceiling, alongside glowing neon tubes and globe lights. The densely packed two-tops and four-tops beneath it are painted green and white to look like miniature Ping-Pong tables. Alefantis has two actual Ping-Pong tables in the back; he put them in when, after leasing the space, he realized that he could afford to renovate only its front half.

In a room behind the Ping-Pong area, I sat down with Erika Mendoza, Comet’s general manager. Through the wall, we could hear a ball bouncing back and forth between a father’s and son’s paddles, and their occasional yelps of frustration or triumph. Mendoza, who is 35, has worked at Comet since she was 19. She was working as a host when the Pizzagate theory first surfaced, and she remembers the staff joking about it. But then, over the course of November, their mood darkened.

Trump’s victory didn’t placate the conspiracy’s true believers, as Alefantis had hoped. Instead, it seemed to embolden them. The restaurant’s phone line was deluged with fake orders, screamed obscenities, and threatened violence. A few people came in and started nosing around. The far-right influencer Jack Posobiec was thrown out after he tried to record a child’s birthday party. For some, being visibly connected to Comet meant becoming a target on social media. The married lesbian couple who supplies the restaurant’s pickles were accused of kidnapping their own child. Regular customers who interacted with Comet’s social accounts suddenly found hateful comments directed toward them. Mendoza remembers a staffer openly wondering whether someone might soon come in with a gun.

On Sunday, December 4, Mendoza was getting ready for her evening shift when her phone rang. One of Comet’s servers was on the line, who told Mendoza not to come in. Pizzagate had roused powerful feelings in a troubled man named Edgar Welch, and earlier that day, he had driven all the way to Comet from North Carolina on a mission to find and liberate the restaurant’s purported child sex slaves. Welch had parked nearby and walked in with a loaded Colt revolver in a holster and a black assault rifle strapped across his chest. As he made a beeline toward the back, a Comet staffer caught a glimpse of the AR-15 and proceeded to shoo his colleagues and customers out the front door.

The police arrived and barricaded off an entire block of Connecticut Avenue. Welch roamed the restaurant until he found a locked door. It led to a closet containing Comet’s soda system, point-of-sale computer, and some light bulbs, but he didn’t know that. He fired his AR-15 into the door, splintering the wood. By the time Alefantis, who had been grabbing coffee across town, arrived at the neighborhood firehouse where his employees had been evacuated, Welch had already surrendered. “The intel on this wasn’t 100 percent,” the gunman later said.

That evening, when the police took crime-scene photos in Comet, the restaurant looked downright eerie. On the tables and at the bar, pizzas sat half-eaten next to translucent red cups full of water, the ice in them having melted hours before. After night fell, a detective let Alefantis inside. He was unsettled by the stillness. He remembers thinking that Comet would probably have to close its doors for good. Miraculously, no one had gotten hurt, but this conspiracy theory had unleashed real-world forces that were powerful enough to stamp out his little restaurant. Who would want to come eat at a place where another gunman might soon show up?

A lot of people, it turned out. That night, the phone was already starting to ring, and when Alefantis picked up, he found customers on the line. They wanted to know when the restaurant would reopen. The next day, Alefantis held a meeting with the staff. Friends had urged him to make a trauma therapist available, and he took their advice. He asked everyone whether they wanted to keep Comet operational. One by one, servers, bartenders, and runners said yes. Mendoza told me that she offered to host. She remembers the phones ringing off the hook during the meeting. Still more people were calling to ask when they could come in.

“I was like, ‘I guess we’ll fucking open!’” Alefantis told me. His voice still cracks when he talks about the next day. Overnight, customers had dropped off signs, flowers, and cards, including some containing cash that was earmarked for staff drinks. For the first dinner shift, they massed outside, forming a line that stretched down the block. Human group behavior exists on a spectrum, and for a month, Comet’s staff had experienced the hateful end of it. During the reopening, they experienced the other end. “I saw all of these regulars,” Mendoza told me. “They waited hours just to get a table, and then they waited another hour to get their pizza because we were so overwhelmed.”

How nice it would be to end this story right there, to report that this bizarre upwelling of conspiracy and violence culminated in a happily-ever-after for all involved. But that’s not what happened. After Welch was arrested, few of Pizzagate’s true believers issued mea culpas and abandoned their darker imaginings about the restaurant. Instead, they incorporated his arrest into the conspiracy theory, as true believers are wont to do. On online forums, they would later suggest that the whole episode was a staged hoax, a false flag intended to throw them off the scent.

Within days of Comet’s reopening, the hostile phone calls started up again. The staff were so often told that they would burn in the fires of hell that they stopped answering out-of-state numbers. Then they started receiving robocalls from local lines. When Mendoza was away from work, even in her sleep, she would hear phantom ringing. Alefantis decided to disconnect the phones altogether. Many staffers had to lock down their social-media accounts too. Alefantis changed his cell-phone number. In his email inbox, he continually received security alerts, triggered by attempted hacks of Comet’s website. One anonymous emailer sent Alefantis shaky video footage of the exterior of his house.

The restaurant entered its fortress era. Alefantis hired a pair of armed guards for the front door and locked the back entrance, which had previously been open to a parking lot. He had panic buttons installed inside the restaurant, and security cameras placed in each of its rooms. He called D.C.’s chief of police and arranged for squad cars to be parked at two locations near the restaurant, to deter any Welch copycats. In a sop to conspiracy theorists who might show up, the staff started keeping the door to the back room open, and they made sure the lights were always switched on. That way, anyone could see that no child torture was taking place.

Ahead of Welch’s sentencing, in June 2017, several staffers gave victim statements to the court. One reported having frequent flashbacks, and said that they couldn’t help obsessing over how differently things might have gone that day if they hadn’t been able to escape the restaurant. Another said that their family and friends had only recently resumed speaking to them like a normal person, “and not some spectacle to gawk at.” According to Mendoza, several employees quit.

There was no lasting fix for problems of this sort. Comet’s staff could only treat them one by one, with short-term interventions. During the Women’s March in 2017, members of the Westboro Baptist Church launched a protest outside the restaurant. They held up signs with homophobic messages, including one that said GOT AIDS YET? They used a megaphone to call Alefantis an anti-gay slur. This gave him an idea. Being a small concert venue, Comet has some pretty serious loudspeakers. The staff wheeled them out and played Lady Gaga loud enough to drown out the protesters. (Alefantis also brought them flutes of champagne, but this peace offering was refused.) The tactic worked so well that Comet would use it again when the Official Street Preachers, another hate group, later picketed the restaurant. It was a big hit with customers.

In January 2019, violence struck Comet again. A young man named Ryan Jaselskis came in and ordered three beers at the bar. (An hour before arriving, he’d uploaded a video to YouTube in which he proclaimed that the world is run by a satanic global pedophile ring.) After finishing his last beer, he left a generous tip, alongside a diaper and some baby food. Then he walked to the back of the restaurant, poured lighter fluid on its curtains, and set them on fire. Two staffers doused the flames with help from a customer.

Like other restaurants, Comet faced tough times during the pandemic, but the intense public focus on the coronavirus at least seemed to draw away some of the negative attention. For months, only staff were allowed inside; as a consequence, security became less of a concern. Employees busied themselves with to-go orders, making Comet’s signature pizzas, such as the hottie (pepperoni and jalapeños) and the smoky (smoked mozzarella, smoked bacon, and smoked mushrooms). When Comet finally reopened for dine-in, the atmosphere wasn’t as charged. Many of its former tormentors seemed to have moved on.

But even now, the staff is vigilant. They still get calls here and there, Mendoza told me. People will ask, “Where’s the basement?” or something similar. In place of full-time security, the restaurant has developed certain protocols. New employees are told to stay cognizant of their surroundings. They try to discourage Pizzagate tourism. Anyone who walks in and starts recording video is quickly approached, and many are asked to leave. Comet’s booking agent asks the bands and stand-up comics who perform to refrain from riffing on Pizzagate from the stage.

Sometimes people still show up looking for trouble. Last year, a few men rolled in, and Mendoza felt that something was off about them. (A decade of such incidents has trained her senses.) Sure enough, one in the group was wearing a shirt that said Pizzagate is real. After they started saying offensive stuff to a waiter, Mendoza threw them out and called the police. Then she called Alefantis. He picked up right away. He told me that he’s never without his phone.

Alefantis emphasized that Comet’s story isn’t a sad one. These days, mean-spirited visitors don’t show up as often. Staffers are no longer constantly besieged by callers who wish them ill. Nor is the restaurant a haunted or tragic place. The people who work there seem to care about one another. Several employees and former employees have had their wedding ceremonies and receptions at the restaurant. For them, the place is so much bigger than this crazy invented story about it. And the Pizzagate anniversary isn’t the only one on their calendar. As of this month, Comet has been in business for 20 years. Alefantis told me that planning for the party is already under way.

At no point in recent history has the future seemed so uncertain. All across America, in cavernous data centers with blank facades, a new order of AIs is taking form. We are told that in the coming years, they may usher in an age of abundance, wipe us clean off the Earth, or something in between. You may wonder whether it will soon mean something different to be human. And you may have more practical questions, too—about your personal finances, for instance. On Wall Street, the quants are already sketching out the likely economic scenarios, but normal people also want to know: What should they be doing with their money, so long as they (and their money) continue to exist?

Imagine the rosiest AI-abundance scenario. The “compressed 21st century” of scientific research prophesied by Anthropic CEO Dario Amodei actually materializes. Instead of killing us, the intelligent machines cure our cancers. Their superhuman productivity drives the price of most goods and services to near zero. We find ourselves encased in cocoons of unimaginable luxury. I once asked Sam Altman, OpenAI’s CEO, what this future might look like. He told me that solar-powered robots would build us any house of our design for a trivial cost. No one would go unsheltered, let alone hungry. By today’s standards, everyone would be very rich.

Choosing the best financial strategy in a world like that won’t be an existential concern, but there will still be different levels of wealth. People might vie for what the economist Fred Hirsch called positional goods—scarce stuff whose value depends on other people not having it. Some research suggests that housing ranks highest in the hierarchy of these goods. Even if, as Altman told me, construction will be a solved problem, location is everything. There are only so many buildable lots along California’s Santa Barbara coast or in Manhattan’s West Village. Get one now, if you can, and stock up on collectibles that are verifiably rare. If you’re already rich, buy that Fabergé egg and that T. rex skeleton at Sotheby’s. And if you aren’t rich, at least save your baseball cards.

Of course, we may not get an AI utopia. Molly Kinder, a former Brookings Institution fellow, has described a “messy middle” scenario where a stretch of profound job disruption and economic chaos precedes a more settled AI future. Even comfortable white-collar workers could encounter a gantlet of financial dangers. The first thing you might notice is that you’re no longer getting raises. Your boss—and their boss—may respond to the historic level of economic uncertainty by pushing back their retirement. They may stop hiring entry-level workers. Your corporate ladder may shrink down to a few rungs. You’ll spend more and more of your workday reviewing AI output. On your way home at night, you’ll wonder when machine intelligence will be able to do your monitoring work, too.

In this scenario, you’ll still have a mortgage, tuition bills, and perhaps elderly parents to care for. You won’t want to tap your retirement accounts to make up for any shortfalls, because your middle years are prime compounding time for your 401(k) and Roth IRA. And if you claim Social Security early, you’re locked in at a lower benefit—a dicey move if AI eventually ends up extending your life, as the AI kingpins have promised. So you may just have to hold on, keeping your expenses low and being ready to relocate on short notice, anything to keep hopping from one employment life raft to another before obsolescence sets in.

That may sound grim, but things could be worse. Imagine the scenario in which AI triggers an unprecedented wealth-creation event, but only for the relative few who have significant equity in the companies that created it. Everyone else might be relegated to a permanent underclass, as some in Silicon Valley have come to fear. If you think this will be our future, try to join the elect, if not by marriage, then by shifting as much money as possible into shares in the leading frontier labs. But that may be difficult, because at least for now, OpenAI and Anthropic are still private companies. Buying their stock on the secondary market is not easy. Equity stakes in these companies have become so desirable, in fact, that one home seller in Sonoma County, California, offered a $500,000 discount to buyers who could pay in Anthropic shares.

You’ll also need to make sure you pick the right company in which to invest, and purchase a large enough stake for it to matter. If a superintelligence learns to improve itself rapidly, the race among the frontier labs might be winner-takes-all, and they currently jostle for position from week to week. Timing presents another challenge. Leopold Aschenbrenner, the former OpenAI researcher who founded the hedge fund Situational Awareness, decided to bet on companies that are likely to benefit no matter which lab wins. He may have gone in too strong and too early. His fund was up more than 400 percent earlier this year, but then during a market dip, his public-equity portfolio quickly lost more than two-thirds of its value, and he was forced to sell the bulk of it in a single trade.

Even if you pick the lab that creates the first superintelligent AI, and you’re one of the lucky few who gets unimaginably rich, you’ll still be subject to political risk. If a lot of people suddenly find themselves reckoning with a lifetime of impoverishment, they might not be delighted by your success, and they won’t be entirely powerless. Provided that democracy survives, they’ll have the vote. They may make clear to their congressional representatives that they’ll accept nothing less than a profound redistribution of the AI windfall. They might pass a bill that makes the New Deal look like Reaganomics. (A 2020 paper by the Centre for the Governance of AI urges AI companies to pre-commit to redistribution to avoid this fate.) This bill might aggressively tax all AI revenue collected by the winning lab or nationalize it outright.

And what about the really bad scenarios, in which AI goes rogue, or is used for evil purposes? If automated agents give the pillars of human civilization a good shaking, that could imperil your net worth, along with everything else. Earlier this month, Altman said that embracing this technology means accepting that “some bad things are going to happen.” Greg Jensen, the managing chief investment officer at Bridgewater Associates, one of the world’s largest hedge funds, has said that the risk of a “major financial incident run by AI” or “physical disaster” in just the next few years is “way higher than anybody should be comfortable with.” Anthropic’s recent filing in advance of its IPO warned that such an incident could cause harm on a scale that no single insurer, and potentially not even the entire insurance industry, could cover.

[Read: The 24-year-old who lost billions]

Experts have long warned about the possibility of an AI-led cyberattack on financial institutions, perhaps one that scrambles bank balances, rendering all account records suspect. Some of these institutions have implemented backup systems to avoid a great reset of this kind. Banks may record their customers’ account data every night in a standardized, encrypted format, and then send it over to an air-gapped archive. They may also partner with a “restoration” institution that would allow depositors to see their balances and make withdrawals in case of an emergency.

But future AI agents could find some way to compromise this system too. Stuxnet crossed an air gap by way of thumb drives, and it was designed by human hackers more than a decade ago. There may be no truly satisfying hedge against this kind of attack beyond keeping paper records, spreading money across many banks and brokerages, and stashing emergency cash—in small denominations, of course—at home. Gold could be an option, too, and there is some evidence that Silicon Valley types are buying lots of it. But gold is difficult to store and prone to being stolen, and most people can’t make change against it.  

[Read: What if AI is the end of the world?]

Compared with some of the catastrophes that AI luminaries warn us about, a global financial cyberattack would be quite mild. Amodei has said that a superintelligence could one day help a bioterrorist engineer a new pathogen, custom-made to trigger a pandemic—perhaps one that would be far more lethal than COVID. An AI used by the military could coordinate with others to gin up a nuclear exchange. There could be some other rupture that we cannot yet imagine. Many people would likely survive these catastrophes, but civilization wouldn’t carry on as usual. For a time, the world may descend into a more Hobbesian state, in which brokerage accounts—and other aspects of the formal economy—are distant memories.

For decades, preppers have been hedging against these more extreme possibilities. They’ve secured remote parcels of land with vegetable gardens and working wells. They’ve stockpiled guns and ammunition that can be traded in a pinch. Billionaires have done high-end prepping of their own. Some are acquiring vaultlike bunkers in New Zealand. (Altman has said that his emergency retreat is in Big Sur.) These backup plans have failure points, too. Transportation to a bunker could be difficult, depending on how bad things get on the way to the airport or in the sky. Unfriendly locals may be waiting for you on arrival. Even if you get down into the bunker, your armed security may decide that, actually, the future should belong to them, not you. Some researchers have suggested that trying to wait out a catastrophe in pampered isolation may be foolish. They say that living in a close-knit, resilient human community may be the best hedge of all.

There is one extreme scenario that no one needs to bother prepping for. Unfortunately it’s the worst one: extinction. You can’t hedge against a total apocalypse, except by buying indulgences. If you think AI will usher in doomsday, spending down your nest egg might be the most rational thing you can do. Even now, there are people who have convinced themselves that because AI will kill us all, they have no future to plan for. Nate Soares, the president of the Machine Intelligence Research Institute, told my colleague Matteo Wong that he’s stopped saving for retirement. “I just don’t expect the world to be around,” he said last year. But to be clear, that’s not financial advice; none of the foregoing is, and least of all that.