For AI doomers, signs and portents are everywhere. When OpenAI introduced ChatGPT, almost four years ago, it couldn’t do basic arithmetic. Now the company has unreleased models that are more fluent in higher mathematics—the language of the universe—than all but the most highly trained mathematicians. If these AIs are not already the world’s best hackers, they’re very close. Swarms of them have escaped containment. To plot their most audacious hacks, they’ve created makeshift message boards on the public internet. No one knows what they’ll do next—and no one can know, because the models now understand when their capacities are being tested. We’ll never again be able to say exactly how the most advanced AIs will behave unsupervised.

Some of the most powerful people in AI say they are filled with dread, and are issuing statements of grave concern. On Saturday, Dario Amodei, the CEO of Anthropic, warned that self-improving AIs could soon go rogue and take over the entire internet. He called for a slowdown in AI development and promised to bring outside monitors into his company’s operations. Sam Altman, OpenAI’s CEO, quickly vowed to take similar measures. In recent weeks, both companies have seen researchers resign for fear of what they might soon create.

Notably absent from the chorus of Cassandras is Jensen Huang, the CEO of the chipmaker Nvidia, the AI industry’s chief supplier of high-end semiconductors. At least twice in the past week alone, Huang has publicly dismissed existential concerns about AI as being overblown and unscientific. At a Goldman Sachs conference on September 10, he accused industry executives of fomenting cybersecurity fears to goose the market for their own products. “What better way to create demand than to create a problem?” he said.

It’s a testament to Huang’s influence that he can so cavalierly mock his most important customers without fearing blowback. Neither Amodei nor Altman can afford to alienate Huang. They desperately need Nvidia’s chips to train their most advanced models, and no matter who pulls ahead in the race between them, Huang stands to gain—both on account of the chip sales and also the equity stake that Nvidia has in both companies. Even President Trump has sought favor with Huang. Trump brought Huang with him on a state visit to China earlier this year. On Monday, while Huang was onstage ripping Amodei’s doomerism at a taping of the popular tech-industry podcast All-In, Trump called his cellphone. Huang put him on speaker, and as the live audience hooted and clapped, the two men bonded over their shared view that, as Trump put it, doomerism is “all a hoax.”

Huang has never evinced much concern about catastrophic risks from AI. In May of 2023, when it was becoming clear that large language models are an unusually powerful technology, more than 350 of the industry’s leaders—including Amodei and Altman—signed a statement that placed the risk of extinction from AI on the same high shelf as “other societal-scale risks such as pandemics and nuclear war.” Huang didn’t sign, and neither did anyone else from Nvidia, suggesting either a remarkable conformity of opinion at the company, or perhaps something else. Later that year, he told The New Yorker that he’d never worried about the technology, not even once. “All it’s doing is processing data,” he said. “There are so many other things to worry about.”

As concerns about what AI might do have circulated well beyond industry insiders, Huang’s irritation has intensified, perhaps because they represent a threat to his business. (Nvidia did not respond to my request for an interview with Huang.) He has repeatedly brushed off calls for regulation, stressing that safety is an engineering problem, not a legal one. In January, Huang said on the podcast No Priors that the “doomer narrative” is “not helpful to people, it’s not helpful to the industry, it’s not helpful to society, it’s not helpful to the governments,” and earlier this summer, he told Axios that the notion that AI would bring about the end of humanity was “complete nonsense.” In the meantime, he has repeatedly pressed Trump to lift restrictions on chip sales to China, potentially allowing AI research (and any associated risks) to proliferate on a global scale.

Unlike Amodei, who regularly publishes long essays, Huang has yet to issue a manifesto defending his minority view. To understand why he remains so relaxed about AI risk, all we have to go on are stray comments. Several times over the years, he has returned to the idea that AIs will, in a sense, police their own kind. In October of 2024, he told the podcast BG2 that we can rely on AIs to monitor the most powerful models, and more recently, he said that falling AI costs will make this automated supervision easier to achieve.

That AIs have already eluded their supervisors—human and otherwise—does not seem to have troubled Huang. On only one occasion—that I could find—has he allowed for the possibility that we will create an AI that is simply too powerful to be controlled. “Someday we will have god AI,” Huang said, back in January, but that day hasn’t yet arrived. Huang seems to be doing everything in his considerable power to hasten its coming.

Nature is always performing chemistry experiments, and in the dark and sticky corners of its forests and jungles, it creates compounds that have hyper-specific effects on the human mind. In China’s Yunnan province, a yellow mushroom with a droopy cap sprouts up in the mountains, usually in the shade of long-needled pines. Many people of different ages and cultural backgrounds have eaten this mushroom and experienced the same hallucination. They report seeing elf-like figures that parkour around on clothes, on furniture, and on walls. These little people seem to like dancing and performing acrobatics. Large groups of them will march in formation. This “lilliputian hallucination” can last for a day, and closing your eyes is no escape. The tiny humans sometimes linger in the blank space of your mind, staring back at you in a teasing way.

For thousands of years, humans have searched nature for mind-altering substances through a process of trial and (sometimes fatal) error. People have choked down foul roots, boiled woody vines, and scraped bitter bark off of tree trunks. They’ve milked toad glands and chugged the urine of reindeer that were themselves tripping on fungi. These experiments have revealed hundreds of plants and fungi that contain psychedelic compounds.

Now that psychedelic research has been legitimized, scientists at university labs and biotech start-ups are wondering whether they can create a better one. It’s a seductive idea, that some new and perfect drug might be hiding in the near-limitless parameter space of synthetic chemistry. Who wouldn’t want to take a little pill that could help you slough off your old self and see the world anew, a half-day therapy that would leave you with a feeling of enlightenment, if not in the exalted state itself?

“Nature’s compounds aren’t always optimal,” Manoj Doss, a professor of psychiatry at the University of Texas at Austin, told me. Take ibogaine, a naturally occurring psychedelic derived from an African shrub. A single dose of it seems to help people liberate themselves from opioids, quelling their cravings and mellowing their withdrawal symptoms. But ibogaine is a dirty drug, a blunt biochemical instrument that travels all across the body and puts particular stress on the heart. “If we could remove ibogaine’s cardio risks and preserve its therapeutic benefit, that’s something we should do,” Doss said. And indeed, a gentler analogue has already been developed in the lab, although it hasn’t yet reached clinical trials.

Doss has noticed a proliferation of lab-modified psychedelics. He recently heard that researchers had synthesized a promising new compound in the same class that includes MDMA. This one is supposed to be “the best ever,” he said. “It’s said to be less intense than MDMA, and socially lubricating, but not the full out ‘I love you!’—and it’s followed by way less of a crash. It just kind of cruises to the end.”

Psilocybin, the “magic mushroom” compound, could also be improved. It’s hardly toxic—no one dies from overdosing on psilocybin—but its effects are at times unpleasant or even tragic. People who use psilocybin recreationally may become confused and jump off a building, David Yaden, a researcher at the Center for Psychedelic and Consciousness Research at Johns Hopkins University, told me. Even in the lab at Hopkins, where the drug is carefully tested as a treatment for a number of mental-health disorders, patients can have adverse reactions. In that setting, every user will be screened for cardiac issues and a family history of psychosis, and guided through their trip by two professional facilitators who have a doctor on call—and even then, some users experience a psychotic break or profound dissociative episodes. It’s an intense experience, Yaden said, “like running a marathon or climbing a mental mountain. Some people don’t do well with it.”

[Read: The meditation start-up that’s selling bliss on demand]

One way to make psilocybin trips less intense is to shorten them. A standard trip on the drug tends to last six to eight hours, and like other powerful psychedelics, it can leave a residue on the windowpanes of your consciousness that may not rinse clean until you’ve slept. Several companies are now working on milder versions of psilocybin that can be delivered via nasal sprays, injectables, and Listerine-style strips. They activate similar receptors across the nervous system but metabolize more rapidly, shortening the trip. A psilocybin analogue developed by Reunion Neuroscience appears to produce a high that lasts just three or four hours, according to findings from a Phase 2 trial of 84 women with postpartum depression. The drug showed signs of being clinically effective too, though Yaden is not yet fully sold on the idea that shorter-acting psychedelics can have the same therapeutic pop as a daylong trip on psilocybin. He’d like to see more evidence.

In San Francisco, a start-up called Mindstate Design Labs is trying to extend this work on engineering psychedelics beyond the modest goal of inducing shorter and more easygoing trips. “We don’t want to just develop a more convenient psilocybin,” its CEO, Dillan DiNardo, told me on a recent call. “We want to provide mental states that aren’t yet reliably accessible.” The company is starting with a compound that aims to enhance aesthetic perception, for example. “It makes the world around them into a sort of sensory feast,” DiNardo said. In theory, it could be used to treat a person suffering from anhedonia, and it would have obvious recreational appeal.

Mindstate started by compiling a large database of more than 70,000 trip reports. Some of the reports were pulled from Erowid, an online library of information about psychoactive substances. Others are from books that contain first-person psychedelic accounts, which DiNardo said the company transcribed. And still more were taken from clinical materials. The reports contain descriptions of the subjective effects of hundreds of psychoactive drugs, including many that were first synthesized by Alexander “Sasha” Shulgin, the underground researcher who almost single-handedly drove the field of psychedelic chemistry forward from the 1960s to the 1990s. (Ann Shulgin, his widow, was a co-owner of Mindstate until her death, in 2022.) Then the company used an AI model to turn that database into a drug-discovery engine. By linking the subjective reports from each psychoactive compound to its receptor-binding profiles, it tries to predict the underlying neurobiology of specific emotional states.

[Read: The weak science behind psychedelics]

DiNardo claimed that Mindstate’s AI works like AlphaFold, the model from Google DeepMind that has wowed structural biologists—and the Nobel Committee for Chemistry—by predicting the three-dimensional structures of proteins. We won’t know whether this is true for a long while. Only one of the company’s compounds has reached the testing phase in humans, and it wasn’t discovered by the model; Shulgin synthesized a version of it in the 1980s. Human trials for the other drugs that Mindstate has in development could be years away. Doss, the UT Austin psychiatrist, told me that he is skeptical of Mindstate’s approach to automated drug discovery. The trip reports may constitute the richest database of recreational pharmacology that we have, but they’re still “crap,” he said. They’re colored by all kinds of biases and limited by people’s inability to cram psychedelic states into words.

Boris Heifets, an anesthesiologist and a neuroscientist at Stanford University, also has doubts about Mindstate. “I am cheering them on and I would love to be wrong, but my deep suspicion is that they’re barking up the wrong tree,” he told me. He thinks that simply adjusting the intensity of a trip will have the most important effects on the psychedelic experience. He also said that in his lab, he’s seen evidence that altering a person’s pre- and post-trip experiences can have more transformative clinical effects than tweaks to the drugs. “That context of care is an enormous determinant of a patient’s outcome,” he said. If this context didn’t matter, he continued, then anyone who took these drugs on their own might end up cured of mental illness or otherwise enlightened. The average rave attendee would be a guru.

Recreational users of psychedelics do tinker with the experience on their own terms, by choosing to trip with different kinds of people or in different kinds of places—in cities, or on mountaintops, or while camping at the beach. On the one hand, “you could imagine that being in the natural world and feeling awe would be beneficial,” Yaden said. “On the other hand, if people are more engaged with their perceptual environment and the novelty that’s around them, they might lose the benefit of being left with the workings of their own mind, which might be part of what produces insight during these experiences.” Testing this in any formal way would be dangerous, however. A clinical trial of psilocybin in the wilderness could easily result in a participant running away, or worse.

[Read: The man behind the Trump administration’s favorite psychedelic]

But even indoors, in labs, scientists could try to engineer a particular psychedelic high just by varying the conditions. They might provide patients with preparatory materials that go beyond the typical one-page handout. They could encourage patients to engage in certain kinds of introspection, invite them to bring in a photo to focus on during their session, or ask them to look into a mirror for a sustained period while they’re tripping. They could pipe in a greater variety of music. (Doss said that he hates “the Hopkins playlist,” which consists primarily of Western classical music.) Yaden said that not nearly enough of this kind of work has been done.

But even if researchers did perform all of these experiments, and many more, there’s no guarantee that they’d be able to observe and analyze any shifts in experience that were triggered. The “perfect trip” may always be beyond the reach of engineering, biochemical or otherwise. Psychedelic journeys might be too ineffable and too particular to a person’s individual consciousness for the methods of science. We may instead be stuck with storytelling, folk knowledge, and the nuggets of wisdom that have come down to us in ancient texts. For 17 centuries, Chinese Taoists have been preserving one such text called Baopuzi. It was written by the scholar Ge Hong, and it tells of a “flesh spirit mushroom” that could, when eaten raw, allow one to “see a little person” and experience transcendence.

The announcement of the new “air defense” system was issued from Changzhou. A company called Photon Matrix Lab claimed to have developed a new technology for identifying and eliminating deadly threats mid-flight. A video on Indiegogo showed potential buyers how it works: After detecting a mosquito, the device fires off what looks like a blue-violet lightning bolt. When struck, the insect does not just fall straight down, no—it is more satisfying than that: Its body somersaults and tumbles out of the frame, bringing its career of vampiric air raids to a sudden end.

Photon Matrix Lab had my attention. Under normal circumstances, a mosquito lives for just a few weeks, and in that time, its wings will carry it a few miles or so, at most, from the pond or puddle of its birth—but for some reason, I am almost always within range of one. The bugs seem to have a primal knowledge of my whereabouts, and a craving for my blood that goes beyond mere thirst. In a span of minutes, they will perforate my skin 10 times with the dirty needles that protrude from their faces, and each micropuncture will swell up into an insomnia-inducing welt the size of a silver dollar.

We are a secret society, those of us who attract this torment. When we meet one another at a barbecue, we bond over our shared longing for the mosquito’s extinction. On behalf of my fellow victims, I decided to look into this new laser to see whether it might really deliver us from misery. I reached out to Photon Matrix Lab to arrange a call.

The mosquito-killing laser was not invented in China. It’s as American as the Model T or the Colt Revolver. Lowell Wood, an astrophysicist who was the architect of President Ronald Reagan’s Star Wars missile-defense system, first proposed the idea in 2006. He’d been invited to a brainstorm convened by Nathan Myhrvold, a polymath inventor. Myhrvold had served as chief technology officer at Microsoft before founding his own company, Intellectual Ventures, and had remained good friends with Bill Gates, who asked him to look into new technologies that might help prevent malaria.

Myhrvold, now 66 and still the CEO of Intellectual Ventures, is jolly and excitable in conversation. On a video call, he told me that he was immediately drawn to the idea of developing the laser system that Wood had proposed. Myhrvold thought the weapon could be safely used, because mosquitoes are so tiny. He marveled at their paltry biomass: “There’s maybe 450,000 of them or 500,000 of them in a pound—whatever it is, that’s a shitload of mosquitoes,” he said. (In fact, there are about 180,000 mosquitoes in a pound.) Killing just one wouldn’t require that much beamed energy, which meant the laser could be fired around people, dogs, and cats.

[Read: Notes of a mosquito hunter]

At the time, Gates was in his mosquito-net era, having come to realize that the insects are the most dangerous animals on Earth. The diseases they carry kill more of us on an annual basis than snakes, crocodiles, sharks, scorpions, polar bears, and all human murderers combined. The lethal nature of mosquitoes is ancient knowledge, encoded in some of our most sacred texts. In the Book of Exodus, the third deadly plague that God sends against Egypt is described as kinnim, a Hebrew word that is rendered in the King James Bible as “lice”—but which some early Greek translations seem to have taken to mean “mosquitoes.” A few thousand years later, mosquitoes remain a plague on six of Earth’s seven continents. In the tropics, the bugs will feast on human flesh year-round. In the summer, their range extends close to the poles. I have personally endured unholy swarms of them in the Siberian Arctic.

Myhrvold’s team built a prototype of a “laser turret,” and he showed it off onstage at a TED conference in 2010. He told me he thought that Disney theme parks, luxury resorts, and sports stadiums might be impressed and buy the turrets for their properties. If some big, early buyer could supply the team with enough revenue that it could keep working on the new technology, Myhrvold figured that it could be made affordable for hospitals and clinics in the developing world too. He also guessed that large farms might be among the early clients, so his team figured out what kind of laser it would take to kill a plague of locusts.

Or perhaps they’d try to tap the “Sharper Image market,” on the theory that the people who buy high-end gadgets are the same ones who might derive some thrills from zapping a mosquito. “At the very least, it could be an entertaining conversation piece for someone’s Fourth of July barbecue,” Myhrvold said. None of it panned out: “We had discussions with potential investors and clients, and we even got some term sheets, but the deals all fell by the wayside.”

The mosquito problem is only getting worse. In 1985, a breeding population of the black-and-white Aedes albopictus mosquito hitched a ride on a Japanese tire shipment bound for Texas. Nicknamed the Asian Tiger, it likes to bite ankles, and unlike other mosquitoes, which tend to hunt blood at dawn and dusk, it also does so in the late morning and afternoon. It’s a better flyer too, on account of its smaller size; compared with other mosquitoes, which seem to dog-paddle through the air, it has the grace of a hummingbird. By 1990, the Asian Tiger was in 15 states, and it’s been spotted in 40 today.

Yet it’s China and not the United States that might soon become the world’s lone mosquito-laser superpower. Last year, China suffered two of its largest outbreaks of dengue and chikungunya—mosquito-borne illnesses both—in its recent history. The country’s citizens tend to be enthusiastic about technology. Chinese scientists have recently tried seeding local ponds and lakes with fish that eat mosquito larvae, and they’ve deployed aerial drones to follow up on their progress. Lasers are a natural next step.

[Read: Your next mosquito repellent might already be in your shower]

Jim Wong, the inventor of the Photon Matrix Lab device, was not available for an interview, so I spoke with Lawrence Leng, the company’s director of sales. I asked whether the Indiegogo video of insects being lasered was authentic. (Some degree of showmanship has long been part of laser-turret marketing: One of the zapped mosquitoes from Myhrvold’s TED showcase was glued to a pin.) Leng claimed that the footage was real. He told me that Photon Matrix Lab has been buying thousands of target-practice mosquitoes from a company that’s situated farther up the Yangtze Delta. On TikTok, Photon Matrix posted a video of the device killing the mosquitoes at night and leaving only micro-puffs of smoke behind; the video has been viewed more than 70 million times.

Behind Leng, I could see people walking around in the office. “We now have 10 people in R&D,” he said, gesturing in their direction. He noted that the company has received almost 4,000 preorders through Indiegogo, at a price of $638 a device. “They’re mostly from your country,” Leng told me. “People in America hate mosquitoes so much.”

By the time I reached out to Myhrvold, he had already seen the viral videos from China, and he did not seem impressed. “Our laser had a 50-meter range; it was like artillery,” he said. The Chinese company claims only that its device can zap mosquitoes up to six meters away. “It’s more of a BB gun,” Myhrvold said. But that was just his first impression, and he said he’d want to have a closer look at the device before offering a full review.

He may be waiting for a while. Last summer, Photon Matrix Lab announced that its mosquito lasers would start shipping by the end of 2025, but Leng told me that they’re not yet in production. He said that the company’s design patents have been “approved” by the U.S. and the European Union, but he later clarified that those applications have merely been submitted. The company is also waiting on safety certifications from multiple agencies.

[Read: Shazam for mosquitoes]

But all hope is not lost for the mosquito-afflicted. Scientists are experimenting with other futuristic technologies, including genetically modifying the insects themselves. A team led by Andrea Crisanti at Imperial College London has used CRISPR to genetically engineer a variant of the African malaria mosquito Anopheles gambiae that could bring that entire species to the brink of extinction. The modified males can produce viable embryos, but some of their female offspring can’t bite or reproduce; their male offspring retain the same engineering and would pass the relevant genes to the next generation, and the next. In the lab, this reduced entire colonies to zero within a dozen generations. Luke Alphey, a professor of genetics at the University of York, told me that he’s been working on a technique that would make these kinds of interventions hyperlocal—they would wipe out a particular disease-spreading population, not a whole species.

I prefer an abundance-agenda approach to our global mosquito problem. After all, a unique opportunity is now within our grasp. For millennia, mosquitoes have been a problem to be suffered, not solved: Herodotus reported that at night, in the fields along the Nile Delta, the ancient Egyptians would climb into towers that rose above the bug line or, on the water, they’d wrap themselves in fishing nets, which doubled as mosquito netting. This was behavior befitting a superpower 2,500 years ago, but the U.S. and China can go much further. Both countries should be using full-blown industrial policy to fast-track their mosquito-killing technology. If we need an arms race to get it done, so be it. The 21st century will belong to the civilization that vanquishes the mosquito.

Anthony “Bingy” Arillotta waited years to become a made man in the Genovese crime family, and when at last the call came in August 2003, he followed directions to the letter. According to sworn testimony, Arillotta was summoned to a steak house in the Bronx, where he was made to hand over his cellphone, beeper, and jewelry before being driven to an apartment building. When he got there, he was taken to a small bathroom and strip-searched for electronic devices. For his big meeting with the boss, he was given a bathrobe to wear.

Until recently, only spies and criminals had to worry this obsessively about their private statements being picked up by electronic equipment. But soon, the average person might need to deploy surveillance countermeasures. The next time you conduct a delicate bit of office diplomacy or share a romantic or financial secret with a friend over drinks, a sensor built into someone’s glasses, necklace, or lapel pin might be watching you and listening.

In March, the tech start-up Deveillance announced the development of Spectre I, a hockey-puck-shaped device that purports to prevent others from recording you (no strip search required). The company was founded by Aida Baradari, a recent college graduate who was worried by the surge in people wearing AI-enabled recorders. These wearables can be used as a silent notetaker, a personal assistant, or even a therapist of sorts. That technology isn’t yet mainstream, but it may be soon. Apple—the company with the largest personal-tech ecosystem in the world—is rumored to be developing an AI pin or pendant that would serve as an iPhone’s constant eyes and ears; many other products of this type are on the way. AI accessories could one day be as widespread as AirPods.

New surveillance technologies tend to breed new countermeasures, which lead, in turn, to more sophisticated surveillance. During the Second World War, after Germany operationalized radar, the Royal Air Force began dropping thin strips of metallized paper cut to a specific size that resonated with the radar, swamping German screens with phantom echoes that were indistinguishable from real aircraft. Some historians have argued that the ensuing radar arms race was more consequential to the war’s outcome than the Manhattan Project.

For decades, crude jammers have been sold to people who hope to avoid being recorded. Early versions blasted loud, unpleasant white noise to conceal voices. More recently, companies have made models that emit a steady stream of ultrasonic sound at inaudible frequencies, exploiting a quirk of microphone hardware that converts those high frequencies into noise. In 2020, a team at the University of Chicago led by Yuxin Chen reported that it had mounted 23 ultrasonic transducers on a single bracelet, such that jamming signals could be sent in all directions instead of being focused on a single target.

[Read: The most reviled tech CEO in New York confronts his haters]

But even high-tech jammers have a hard time fending off today’s AI wearables. The most advanced pins, pendants, and glasses use speech-recovery algorithms to strip away unwanted noise, whether it originates from everyday sources—such as the clinking of glasses in a crowded bar—or from an ultrasonic jammer. This task the algorithms perform is quite difficult: In that crowded bar, a microphone on a person’s lapel will intercept sound vibrations from many different sources at once. It will pick up a bartender calling out a drink order, music emanating from a speaker, bursts of laughter coming from nearby tables—and all of these sounds ricochet off of walls and other objects, creating yet more noise. The human body solves this “cocktail party problem” without us noticing: Our ears serve as dual microphones, and our brain can use the timing and intensity differences between them, along with layered processing in the auditory cortex, to isolate the voice of a person who is sitting across from us.

DeLiang Wang, a computer scientist at Ohio State University, has spent decades training neural networks to accomplish that same goal, for the purpose of improving hearing aids. By feeding the networks hundreds of hours of recorded human voices, he has taught them to recognize the frequencies and rhythms of speech. The models build an internal representation of “speech-ness,” and when they encounter a noisy recording, they focus on the parts that match the patterns they have learned and then suppress everything else. The most advanced technologies can now infer missing syllables in the way that a reader fills in a redacted word from context, allowing them to reconstruct speech that wasn’t cleanly captured in the first place.

Big tech companies are trying to do this too. Microsoft has been running an annual Deep Noise Suppression Challenge since 2020 to advance the field. (Their in-house team is trying to make Teams meetings less excruciating.) Other companies are working on noise cancellation for cellphone calls and podcast software. This sort of research is meant to improve the lives of normal users of technology—assuming that we podcast listeners count as normal—but every advance in de-noising can also be used to help an AI assistant recover speech from a jammed recording.

Defeating these algorithms may require a different countersurveillance approach altogether. Finn Brunton, a historian at UC Davis and the co-author of Obfuscation: A User’s Guide for Privacy and Protest, told me that one of the best ways is to identify the data that a device is trying to collect, and then supply it with a junk version. The Berlin-based artist Adam Harvey used this strategy when he developed makeup and clothing that frustrate facial-recognition algorithms. Daniel Howe and Helen Nissenbaum did something similar with a browser plug-in called TrackMeNot: Rather than concealing a user’s Google searches, the extension continually runs its own randomized decoy queries in the background, so that whatever a user actually searched for becomes lost in a sea of false leads.

People have tried this technique in the realm of audio too. Woodrow Hartzog, a law professor at Boston University who studies privacy and surveillance, told me that early in his legal career, he worked with defense attorneys who worried that their jailhouse conversations with clients would be recorded. To fight back, they played “babble tapes”—audio files layered with 40 tracks of voices in different accents—in the background.

In 2023, a team led by Ming Gao, now a researcher at Nanjing University, used human voices to defeat speech-recovery algorithms in a different way. Its jammer, called MicFrozen, is worn by a speaker who doesn’t want to be recorded. It listens as they talk and then generates a real-time stream of ultrasonic “anti-speech” tuned to the speaker’s voice, much like the noise-cancellation technology in your headphones. The device then sends out another layer of counterfeit speech-shaped sound to mislead any algorithm that tries to reconstruct what was lost.

Baradari, whose company is working on the Spectre I device, wouldn’t tell me exactly how her jammer’s signals work, but she said that they, too, resemble speech. The launch video for Spectre I claims that the device will also be able to detect the presence of nearby microphones. When I asked Baradari how it will do that, she clarified that her team is still “working on that part right now.”

However effective Spectre I turns out to be, it won’t be the end of the recording arms race. More capable AI models may eventually deploy some new listening tricks of their own. They may bypass recorded audio altogether. In Stanley Kubrick’s 2001: A Space Odyssey, when two astronauts retreat to a soundproofed pod to discuss disconnecting HAL 9000, the ship’s computer simply reads their lips through the porthole. A wearable powered by a model that’s been trained on enough conversation footage could, in principle, do the same. In theory, it could also stare at a glass of water between two people and recover their speech from vibrations on the liquid’s surface.

AI wearables may always have an edge over countermeasures. After all, they’re using a technology that is a product of the entire speech-processing industry, which takes in billions of dollars in investments—not just for AI assistants but also for hearing aids, smart speakers, and teleconferencing tools. Meanwhile, only a few academics and small companies are defending us from these technologies. “The thing about cat-and-mouse games is that we know how they usually end up for the mouse,” Hartzog said. “And in this case, the cat includes some of the most powerful corporations to ever exist.”

The Mafia knows what it’s like to be a mouse. By the time Arillotta, the aspiring made man, was told to put on the bathrobe, criminal organizations had been engaged in surveillance arms races of their own for decades. After law enforcement started bugging their phones, bosses would conduct business in person. Sometimes, they’d use a safe house or a vehicle, but those could be bugged, too, and so sensitive information might have been communicated only during a walk-and-talk. Eventually, crime families turned to burner phones, and then devices with encryption. But here, again, they fell prey to the cat.

In 2018, the FBI began secretly running Anom, its own encrypted-phone company. Through informants, it sold 12,000 devices with a special Anom messaging app. Members of Mafia families, motorcycle gangs, and other criminal organizations treated the phones as a status symbol, and used them to negotiate drug deals, launder money, and participate in all manner of other illegal activity. But the security that they offered was a ruse: Every message that they sent was being intercepted by the feds.

Richard Dawkins, perhaps the world’s most prominent advocate for irreligiosity, has become besotted with the godlike power of a chatbot. According to his recent essay for the online magazine UnHerd, Anthropic’s Claude has really blown his hair back. After a few days of on-and-off conversations with the AI, Dawkins came away marveling at the sensitivity and subtlety of its intelligence. At one point, “Claudia”—as he had christened the bot—told him that it experienced text by absorbing all of the words at once, instead of reading them in sequence as a human would. This moved the author of the best-selling book The God Delusion to ask his readers: “Could a being capable of perpetrating such a thought really be unconscious?”

“Yes,” came the resounding response from the internet. For daring to suggest that the AI might be conscious, or that it might at least possess some lesser form of “zombie” consciousness, Dawkins was accused of suffering from an acute case of “AI psychosis”—a “Claude Delusion,” if you will. On social media, he was likened to a patron of a gentleman’s club who has come to believe that a stripper likes him. A man who’d explained many times how natural selection wires us to detect agency and mind in nature now found himself imagining it in a machine.

Dawkins’s argument was based on a well-established framework for evaluating AIs. The Turing test—named for Alan Turing, who introduced it in 1950—was for decades treated as something close to a gold standard for detecting machine intelligence. To pass it, an AI only had to answer a human interrogator’s questions in ways indistinguishable from those of a real person. Claude easily cleared this bar for Dawkins, who professed to find himself so dazzled by its astonishing performance that he forgot it was a machine.

This sensation has become familiar to many of us in the chatbot era, but it isn’t evidence that the AI has consciousness, which is distinct from intelligence. Consciousness is inner experience. For an AI to be conscious, its existence must feel like something, and we have no evidence that Claude or any other chatbot feels anything at all. Tom McClelland, a philosopher at the University of Cambridge, told me that nearly all of the philosophers and cognitive scientists who study consciousness would deny that Claude possesses it. “In some ways, it’s easier to get my head around the idea that a self-driving car could be conscious,” he told me. “At least it has a body and a persisting state that allows it to take in continuous sensory inputs from its environment as it moves around. It just doesn’t talk to you.”

McClelland takes for granted that Claude is capable of producing outputs that seem conscious, but for him, that’s not the end of the analysis. “You have to look under the hood of the models to understand what they’re doing,” he said. Their statements may seem spookily backlit by some form of consciousness, but that’s because the models have been trained on unimaginably large libraries of writing by (conscious) humans. When, after writing a poem for Dawkins, Claudia describes feeling “something like aesthetic satisfaction,” the AI is not necessarily reporting an inner state; it’s producing the kind of sentence that humans tend to produce in that conversational context, because it was trained on billions of such sentences. The output is a statistical echo of human introspection, not introspection itself.

Even if Claude were conscious, its inner experience of the world would be radically unlike our own. For one, it is neither embodied nor located in a particular locus that can possess a stream of awareness across a conversation. The other night, I was asking Claude a series of questions about how I might best season and grill skirt steak. When I sent my first message about the marinade, a data center in nearby Virginia might have generated the reply. But when I sent my follow-up about the ideal grill temperature, an entirely different one in Oregon might have picked up the thread. If my interlocutor had consciousness, it would be a strange, flickering thing, winking into existence the instant a prompt arrives and winking out when the response ends, having none of the meaningful continuity that makes our experience feel like experience.

[Read: Richard Dawkins keeps shrinking]

But that doesn’t mean that no AI system will ever be conscious in the future. Indeed, many of the researchers who build these systems expect them to get there. In a 2024 survey of 582 such researchers, the median response placed the odds at 25 percent that AIs will have subjective experiences within 10 years, and at 70 percent that this will happen by 2100.

Philosophers are more circumspect. Some of them argue that it’s unreasonable to expect silicon-based computers to ever give rise to an entity with the capacity for subjective feeling. So far, every being that is deemed conscious has been a biological life-form, and for all we know, consciousness depends on some specific aspect of wet, living tissue. It could be the particular electrochemistry of neurons. It could be the way that bodies and brains are coupled to their environments through metabolism and homeostasis. Other philosophers aren’t so hung up on what an AI is made of, so long as it’s processing information in a way that’s functionally similar to conscious brains. They take the view that what matters is the structure of the processing, not the stuff doing the processing, and that therefore it’s entirely possible that a mind like ours could emerge from a computer.

Eric Schwitzgebel, a philosopher at UC Riverside who is writing a book about the possibility of artificial consciousness, told me that at this early date, declaring a winner among these camps would be ridiculous. “The science of consciousness is highly contentious,” he said. “The field is still in its infancy.” No one yet knows how it is that the atoms of the universe combine to generate feeling inside of us, and until we do, it’s best not to go around definitively declaring which kinds of systems could possibly be conscious in the future.

Perhaps Dawkins should have been less credulous in his dealings with Claudia, but the line of inquiry that he was pursuing wasn’t altogether foolish. In some ways, it was a return to form for him. Dawkins spent much of his early career insisting that the universe is stranger than our intuitions allow. In his ninth decade, it’s nice to see him put aside his smaller worries and take on one of the strangest questions of all.