Anne Neuberger, the Biden administration’s deputy national security adviser for cyber, tells WIRED about emerging cybersecurity threats—and what the US plans to do about them.

Anne Neuberger, a Top White House Cyber Official, Sees the 'Promise and Peril' in AI

We tested the end-to-end encrypted messenger’s new feature aimed at addressing critics’ most persistent complaint. Here’s how it works.

For nearly a decade, cybersecurity professionals and privacy advocates have recommended the end-to-end encrypted communications app Signal as the gold standard for truly private digital communications. Using it, however, has paradoxically required exposing one particular piece of private information to everyone you text or call: a phone number. Now, that's finally changing.

Today, Signal launched the rollout in beta of a long-awaited set of features it's describing simply as “phone number privacy.” Those features, which WIRED has tested, are designed to allow users to conceal their phone numbers as they communicate on the app and instead share a username as a less-sensitive method of connecting with one another. Rather than give your phone number to other Signal contacts as the identifier they use to begin a conversation with you, in other words, you can now choose to be discoverable via a chosen handle—or even to prevent anyone who does have your phone number from finding you on Signal.

You’ll now be able to set a unique username as a way for people to find you on Signal in addition to—or instead of—your phone number.

Courtesy of Signal

The use of phone numbers has long been perhaps the most persistent criticism of Signal's design. These new privacy protections finally offer a fix, says Meredith Whittaker, Signal's president. “We want to build a communications app that everyone in the world can easily use to connect with anyone else _privately_. That ‘privately’ is really in bold, underlined, in italics,” Whittaker tells WIRED. “So we're extremely sympathetic to people who might be using Signal in high-risk environments who say, 'The phone number is really sensitive information, and I don't feel comfortable having that disseminated broadly.'”

In the new features—which are available in beta now, but which Signal plans to roll out in a more final version in the coming weeks—Signal has made three changes, one setting that's now switched on by default and two that are opt-in features. First, by default, your phone number will no longer be visible in your Signal profile unless someone already has the number saved in their phone's address book. Second, you can now choose to create and share a unique username, or a QR code that contains it, with anyone you want to connect with. Mine, for instance, is Andy.01. (Once someone does start messaging you, a little confusingly, they'll see your chosen profile name instead of that username. That profile name, just as before in Signal, doesn't have to be unique, and the person you're interacting with can also change it in their own view of you in the app.)

New settings in Signal allow you to change both the visibility and discoverability of your phone number. Your number’s visibility will be turned off by default. Turning off its discoverability, a more drastic measure, will be left as an opt-in setting for higher risk users.

Courtesy of Signal

The third new feature, which is not enabled by default and which Signal recommends mainly for high-risk users, allows you to turn off not just your number's visibility but its discoverability. That means no one can find you in Signal unless they have your username, even if they already know your number or have it saved in their address book. That extra safeguard might be important if you don't want anyone to be able to tie your Signal profile to your phone number, but it will also make it significantly harder for people who know you to find you on Signal.

The new phone number protections should now make it possible to use Signal to communicate with untrusted people in ways that would have previously presented serious privacy risks. A reporter can now post a Signal username on a social media profile to allow sources to send encrypted tips, for instance, without also sharing a number that allows strangers to call their cell phone in the middle of the night. An activist can discreetly join an organizing group without broadcasting their personal number to people in the group they don't know.

In the past, using Signal without exposing a private number in either of those situations would have required setting up a new Signal number on a burner phone—a difficult privacy challenge for people in many countries that require identification to buy a SIM card—or with a service like Google Voice. Now you can simply set a username instead, which can be changed or deleted at any time. (Any conversations you've started with the old username will switch over to the new one.) To avoid storing even those usernames, Signal is also using a cryptographic function called a Ristretto hash, which allows it to instead store a list of unique strings of characters that encode those handles.

Amid these new features designed to calibrate exactly who can learn your phone number, however, one key role for that number hasn't changed: There's still no way to avoid sharing your phone number with Signal itself when you register. The fact that this requirement persists even after Signal's upgrade will no doubt rankle some critics who have pushed Signal's developers to better cater to users seeking more complete anonymity, such that even Signal's own staff can't see a phone number that might identify users or hand that number over to a surveillance agency wielding a court order.

Whittaker says that, for better or worse, a phone number remains a necessary requisite as the identifier Signal privately collects from its users. That's partly because it prevents spammers from creating endless accounts since phone numbers are scarce. Phone numbers are also what allow anyone to install Signal and have it immediately populate with contacts from their address book, a key element of its usability.

In fact, designing a system that prevents spam accounts and imports the user's address book _without_ requiring a phone number is “a deceptively hard problem,” says Whittaker. “Spam prevention and actually being able to connect with your social graph on a communications app—those are existential concerns,” she says. “That's the reason that you still need a phone number to register, because we still need a thing that does that work.”

The continued phone number requirement means Signal's privacy upgrade is a compromise, says Matthew Green, a professor of cryptography and computer science at Johns Hopkins University who has in the past consulted for both Google and Facebook in their implementation of Signal's open source encryption protocol. “It's a half solution,” says Green. “It's not a perfect solution.”

Green notes, however, that even if it doesn't satisfy the most die-hard privacy advocates, it represents a significant improvement for a much larger portion of Signal's hundreds of millions of users. “There’s a legitimate community of people who wanted to use Signal without giving other people their phone numbers, and they’re going to be very happy with this change. And then there's a more hardcore set of people who don’t want to ever give their number to Signal,” Green says. “I think getting a big set of people serviced is the right direction, and working on satisfying all the other people is something for Signal to keep working on.”

Signal doesn't currently have any road map toward dropping its use of phone numbers as a registration mechanism, Whittaker concedes—she says for now, there's no alternative that wouldn't sacrifice Signal's usability, which she argues would represent a net loss for privacy advocates. But she says that the new phone number privacy features are nonetheless Signal's careful attempt to solve the problem phone numbers represent without losing the qualities that have made Signal popular in the first place.

“It's really about staying true to our principles,” Whittaker says. “In more and more ways—in better and better ways—to fill that promise of easy, usable, private communications.”

_Correction: 2/20/24, 1:25 pm EST: Meredith Whittaker's professional title is president of the Signal Foundation._

Signal Finally Rolls Out Usernames, So You Can Keep Your Phone Number Private

LockBit’s website, infrastructure, and data have been seized by law enforcement—striking a huge blow against one of the world’s most prolific ransomware groups.

For the past four years, the LockBit ransomware group has been on an unrelenting rampage, hacking into thousands of businesses, schools, medical facilities, and governments around the world—and making millions in the process. A children’s hospital, Boeing, the UK’s Royal Mail, and sandwich chain Subway have all been recent victims.

But LockBit’s hacking campaign has come to a juddering halt. A sweeping law enforcement operation, led by police at the UK’s National Crime Agency (NCA) and involving investigators from 10 forces around the world, has infiltrated the ransomware group and taken its systems offline.

Graeme Biggar, the director general of the NCA, says the group has been “fundamentally disrupted.” The law enforcement operation, called Operation Cronos, has taken control of LockBit's infrastructure and administration system, seized its dark-web leak site, accessed its source code, seized around 11,000 domains and servers, and obtained details of the group's members. “As of today, LockBit is effectively redundant,” Biggar said at a press conference in London, appearing with law enforcement officials from the FBI and Europol. “We have hacked the hackers,” he says.

The action is one of the largest and potentially most significant ever taken against a cybercrime group. Biggar says the law enforcement officials consider LockBit, which is global in nature, to have been the “most prolific and harmful” ransomware group that has been active in recent years. It was responsible for 25 percent of attacks in the past year. “LockBit ransomware has caused losses of billions,” Biggar says of the overall costs of attacks and recovery.

In addition to the seizing of technical infrastructure, the law enforcement operations around LockBit also include arrests in Poland, Ukraine, and the United States, as well as sanctions for two alleged members of the group who are based in Russia. The group has members spread around the world, the officials said.

Nicole Argentieri, acting assistant attorney general at the US Department of Justice, says LockBit has received more than $120 million in ransomware payments, and that the action announced against the group is just the start of the clampdowns.

The law enforcement action against LockBit was first revealed when its ransomware website dropped offline on February 19 and was replaced by a holding page saying it had been seized by police. The LockBit group, which debuted as “ABCD” before changing its name, first appeared at the end of 2019. Since then, LockBit has rapidly attacked businesses and grown its name recognition within the cybercrime ecosystem. “LockBit has been a thorn in the side of businesses and governments for years, with well over 3,000 publicly known victims, and \[has been\] seemingly untouchable,” says Allan Liska, an analyst specializing in ransomware for cybersecurity firm Recorded Future. Lockbit’s long roster of victims include various US government organizations, ports, and automotive companies.

LockBit operates as a ransomware-as-a-service operation, with a core handful of members creating its malware and running its website and infrastructure. This core group licenses its code to “affiliates,” who launch attacks against companies, steal their data, and try to extort money from them. “LockBit is the last of the ‘open affiliate’ ransomware-as-a-service offerings, meaning anyone willing to cough up the cash can join their program with little or no vetting,” Liska says. “They likely have had hundreds of affiliates over the course of their run.”

The Notorious Lockbit Ransomware Gang Has Been Disrupted by Law Enforcement

The widespread use of mines has left Ukrainians scrambling to find ways to clear the explosives. New efforts to develop mine-clearing technology may help them push back Russia's invading forces.

Oleksandr Kryvtsov had enough.

The owner of an agricultural company in Hrakove, near Kharkiv, Kryvtsov found his land littered with land mines. That region of Ukraine, occupied by Russian forces for nearly eight months, had been pockmarked with explosive ordinances. The threat meant that farmers like Kryvtsov had to let their fields lay fallow. Even though Kryvstov’s fields were once part of Europe’s breadbasket, Ukraine’s mine clearance teams were overworked and under-resourced.

So Kryvtsov came up with his own solution. He jimmyrigged a plow onto an old tractor, with massive steel rollers underneath. On the side, he painted the yellow and blue Ukrainian flag. Kryvtsov connected a remote-control steering system and, from afar, he drove his Mad Max-style tractor over his fields, detonating any mines lurking under the soil.

The makeshift operation has worked well, Kryvtsov told Reuters, even clearing an anti-tank mine.

Kryvstov’s story is an example of incredible Ukrainian ingenuity—a nation of gilders, working to invent, adapt, and repurpose technology to defend themselves against a better-resourced, larger, determined enemy. But it’s also an ominous sign of just how bad the problem is.

In recent months, WIRED has investigated the technological challenges and opportunities facing Ukraine as it tries to defend itself and recapture its territory. One particular problem, unsung by the Western media but frequently cited by Ukrainian officials, are the haphazard minefields across Eastern Ukraine.

WIRED has spoken to a range of engineers, government officials, and humanitarian mine-clearance experts, and consulted Ukraine’s new mine clearance plan. It is apparent that Kyiv is prioritizing the problem, but without a significant new influx of money, personnel, and technology, the threat of these mines could hobble Ukraine’s economy, frustrate future counteroffensives, and pose a humanitarian crisis for decades to come.

A Humanitarian Crisis, an Economic Cost

Ukraine’s mine problem has been acute for a decade. The full-scale war with Russia has only made it worse. From 2014, when Russia first invaded, to the end of 2021, the United Nations says 312 Ukraines were killed by land mines. Since Russia’s February 2022 invasion, Ukraine has recorded at least 269 civilian casualties, including 14 children. Prime Minister Denys Shmyhal has taken to calling Eastern Ukraine “the largest minefield in the world.”

Those casualty figures only capture the deaths on territory currently held by Ukraine. Behind the front lines, in the Russian-occupied regions of Eastern Ukraine, at least a hundred more have reportedly been killed.

“Twenty percent of the whole territory is dangerous,” Ihor Bezkaravainyi, Ukraine’s deputy minister of finance, tells WIRED. “Right now we’re talking about 150,000 square kilometers.” (The total area, including water littered with naval mines, is nearly 175,000 km².)

Bezkaravainyi is a veteran of the war in Eastern Ukraine—he lost a leg to an anti-tank mine in 2016. He’s now responsible for coordinating the mine-clearance effort behind the front lines, giving Ukrainians back their property and recovering damaged agricultural lands. It’s not an easy task.

“It looks like the _zone rogue_ in France after World War One,” Bezkaravainyi says, referring to the areas near Germany and Belgium that remain contaminated by land mines to this day.

Conducting surveys to identify those mines will be a gargantuan challenge. Actually clearing them will be even more taxing.

Russia has deployed older anti-tank and anti-personnel mines—of the kind the world has ample experience dealing with. But it is believed that this is the first time the sophisticated PTKM-1R anti-tank mine, which detonates only when it picks up a certain seismic signature, has been used in battle. Russia has also made liberal use of the more advanced PFM-1 mine, also known as the “butterfly mine,” made mostly from plastic and liquid explosive. These mines are particularly odious because they can be scattered in huge quantities from afar or from the air, meaning that they are impossible to track. Because they are colorful and plastic, they can be mistaken by children as toys.

Beyond purpose-built mines, Russia has also littered Ukraine with unexploded munitions and “improvised explosive devices and booby traps,” according to a draft version of Kyiv’s plan to decontaminate the country, prepared late last year and provided to WIRED.

Until now, Ukraine has not had a national plan on how to deal with the mine problem—its ad hoc response has been split between the military, NGOs, a small number of private mine-clearance companies, and a small network of government mine-clearance operators.

In 2021, before Russia’s full-scale invasion, Ukraine had certified just four “mine action operators” to conduct the mine identification and clearance. Since the start of the war, that number has grown to “only 23.” That number is simply “not adequate,” the plan states.

This National Mine Action Strategy was devised to bring consistency and focus to this effort. But it warns that the scope of this problem “cannot be solved in a short-term perspective.” Kyiv hopes it can assess the entirety of its lands, to identify which areas are actually contaminated and which are safe to use, by 2029. By 2033, Ukraine aspires to have decontaminated 80 percent of its previously occupied territory. The strategy does not provide a date for when the whole country might be free of mines.

If Ukraine wants to meet these goals, it will need significantly more staff, technology, and equipment than it has now.

This will be a tough hill to climb. Humanitarian demining groups are spread thin across many global conflict zones, while commercial operators tend to be prohibitively expensive.

This is the reality that pushed Kryvtsov, the farmer, to take matters into his own hands. But Kyiv warns that these “black sappers”—unlicensed and unsanctioned mine-clearance operations—are dangerous and unreliable. Still, the government recognizes that, unless it can dramatically scale up its own operations, these freelance mine operators will become a popular choice for locals frustrated by the slow pace of progress.

Like many aspects of Ukraine’s war effort, Kyiv believes the solution is at home. The strategy calls for a substantial investment in Ukraine’s industrial capacity to produce mine-clearing equipment, research demining technology, and train demining teams. The World Bank estimated in early 2023 that the total cost of identifying and clearing these mines would be nearly $38 billion. Kyiv expects the true cost will be higher.

A Military Challenge

In the snow-covered fields near the front lines, the Ukrainian Armed Forces have been deploying autonomous demining vehicles—which, although purpose-built and donated by European allies, look an awful lot like Kryvstov’s homemade version.

Since the start of the war, these military demining teams have cleared more than 280,000 mines—at a pace of more than 2,200 every week. Its work is entirely separate from the humanitarian teams run under Bezkaravainyi's department.

The military may have cleared a staggering volume, but the work is impeded by a lack of equipment. The military boasts 262 separate demining teams, but it has just six demining vehicles.

In an essay for _The Economist,_ former commander-in-chief of the Ukrainian Armed Forces Valerii Zaluzhnyi wrote that his forces were initially relying on “technically outdated pieces of equipment” to conduct this operational mine clearance. With Western donations, “it was possible to slightly augment the capabilities of engineer units … but given the unprecedented scale of these barriers, even such capabilities are objectively lacking.”

Much of the analysis of Ukraine’s failed summer counteroffensive has focused on the offensive gear it lacks—artillery shells, fighter jets, drones, and long-range missiles. But even if it had managed to pierce the Russian front line, Ukraine faced layers of other defensive structures, including between 15 and 20 kilometers of minefields.

As Zaluzhnyi notes, Russian reconnaissance drones have kept a watchful eye on these minefields, targeting any Ukrainian teams dispatched to clear them. “In case of successful mine barriers breaching, the enemy quickly restores minefields in these areas,” he wrote.

European allies, in particular, have donated mine clearance vehicles—including retrofitted German-made Leopard 2 tanks—but they have been hard hit by Russian forces.

While clearing fields near the front lines is difficult, risky work, Russia is capable of laying these fields remotely and quickly. The ISDM Zemledeliye, a mobile mine-laying system that sits on the back of a truck, can carry 50 rockets, each filled with anti-personnel or anti-tank mines that scatter over a targeted area. The system allows the operator to lay minefields from as far as 15 kilometers away. One pro-Kremlin Russian media outlet recently remarked that the Zemledeliye, which translates to “agriculture” in Russian, “sowed” the defeat of the Ukrainian summer counteroffensive.

“The Ukrainians didn't necessarily have the equipment, the type of trained brigades, etc, to break through that defense and overcome the Russians, who were defending in a doctrinally consistent and actually quite sound way,” Karolina Hird, an analyst at the Institute for the Study of War and the deputy team lead for their Russia desk, tells WIRED. “Which brings us to where we are today.”

Even if Kyiv manages to overcome all of those other problems, if it cannot figure out how to clear the Russian-laid minefields, its progress risks being squandered.

“One of the big operational problems is, how do you increase by an order of magnitude the detection of mines, mapping of minefields, and the clearance of them—whilst denying the Russians visibility of them?” Mick Ryan, a 35-year veteran of the Australian Army who has traveled to Ukraine frequently during the war, tells WIRED. “And these are pretty significant problems, but they’re known problems, right?”

Ryan says there needs to be a deeper recalibration of the relationship between Ukraine and NATO. At the beginning of the war, the transfer of knowledge and expertise from NATO to Ukraine may have been largely one-directional, but today, Ukraine’s expertise in modern warfare certainly rivals many of its benefactors.

“Ukrainians and NATO, they just need to divide up the problems and solve them,” Ryan says. “I mean, this isn't inventing the nuclear bomb.”

The Technology

As the National Mine Action Strategy notes, research on mine clearance has been sorely lacking.

There has been, the strategy says, a “lack of systematic and centralized work on the introduction of innovative technologies in the field of Mine Action, in particular, unmanned aerial vehicles, the use of satellite images, artificial intelligence, data collection and analysis systems.”

It’s a frustration that Federica Mezzani knows well. Since 2019, she’s been researching how new technologies can help improve mine detection strategies—but it is a field, she says, which had been “completely forgotten.”

Despite the fact that an estimated 110 million mines are still active around the world, they are primarily distributed in poor and war-torn countries. While NGOs such as the HALO Trust have worked to steadily decontaminate those territories, the research and development has been piecemeal and slow. It simply hasn’t been a priority.

But Mezzani, along with her colleagues in the Department of Mechanical and Aerospace Engineering at Sapienza University in Rome, set out to prove that new technology could help with this old problem. There had been some research testing how drones could be used to identify unexploded ordinances, but not much. Mezzani wanted to take it a step further, dispatching drone swarms equipped with ground-penetrating radar to methodically scan each section of the ground, the way a human team might. Algorithms could essentially automate mine detection, she believed.

In a series of small-scale experiments, Mezzani’s technique worked.

“The experimental campaign proved the effectiveness of the algorithm, which appears as a powerful tool to automatically detect buried objects with even small metal content,” reads her paper, published in _Advances in Nonlinear Dynamics_ in 2022.

“The technology is ready,” Mezzani tells WIRED. “I think that it's been ready for many years, actually.”

When the full-scale war began, research efforts like Mezzani’s were few and far between. That meant figuring out these strategies from the ground up.

Part of the challenge is about confidence. As Bezkaravainyi explains, humanitarian mine clearance operates on a zero-tolerance policy for civilian deaths—if they mark a territory as uncontaminated, they must be absolutely sure it is entirely safe. Where possible, that also means defusing the land mines instead of exploding them and further contaminating the soil.

This process is significantly slower than how the military clears territory. Prioritizing speed, the army may blast a path through an active minefield in order to advance quickly without fully clearing it. To that end, humanitarian mine clearance operates on the Swiss cheese model: applying multiple imperfect strategies on top of each other.

Bezkaravainyi explains that their process normally involves consulting high-resolution satellite imagery of the territory and identifying land mines from the sky. From there, drones may be dispatched to confirm those locations and identify mines that may be buried or tough to spot. After that, teams are dispatched to sweep the territory.

Last fall, at an international conference on Ukraine’s demining efforts held in Zagreb, Bezkaravainyi’s department unveiled a prototype, developed by American surveillance technology giant Palantir, which used artificial intelligence to help inform how Kyiv approaches mine clearance.

This multilayered approach is increasingly necessary. Magnetometers and thermal scanners, which identify mines by identifying the metal amidst the organic material, were once the gold standard for mine identification. Some mines have electromagnetic shields, protecting them from ground-penetrating radar. The PFM-1 mine, in particular, contains very little metal, making it difficult to detect.

This problem is mostly, but not entirely, of Russia’s making. Reports suggest that Ukraine has also deployed these PFM-1 mines against Russian forces in Eastern Ukraine.

Difficult terrain, such as forests or mud, makes this work more difficult. Ukraine has difficult terrain in spades: It even has a word, _bezdorizhzhya_, for the mud that covers the eastern part of the country in the spring.

“If all the technologies in the world were given to Ukraine, it would not be enough,” Bezkaravainyi says.

An Opportunity

Ukraine is not merely contemplating how to buy and acquire enough technology to do this job—they are developing a plan to become a world leader on mine clearance.

That sort of focus has been sorely lacking for decades. “It’s a type of research that doesn't bring in profits,” Mezzani says. It’s a problem she crashed into during her own research project. “I wouldn't say that we have a technological issue. We have a willingness issue.”

Some 70 countries worldwide are still contaminated by land mines, according to the United Nations, and they kill or maim thousands every year. Most, however, are located in the Global South. The conflict in Ukraine may finally be the impetus to develop the technology and expertise to address that problem.

Indeed, Bezkaravainyi says his department has fielded plenty of offers from companies professing expertise in mine clearance, but many have been unreliable, haven’t delivered, or were outright scams.

If Ukraine can develop both the technology and the industry to do this work, it could provide a critical advantage in the war, boost its battered economy, and provide an enormous service to the entire world.

Brave1, a platform launched by the Ukrainian government to identify innovative projects and connect them to public and private financing, has identified mine clearance as one of its main priorities. Thus far, 30 projects—which range from autonomous land vehicles to more sophisticated detection systems—are part of the Brave1 platform.

If the Ukrainian government can spur the creation of a domestic demining industry, it will speed up its economic recovery, help war-torn countries the world over, and maybe even help win the war.

The Danger Lurking Just Below Ukraine's Surface

Plus: State-backed hackers test out generative AI, the US takes down a major Russian military botnet, and 100 hospitals in Romania go offline amid a major ransomware attack.

While the response to Cowles’ tale has been a mix of praise and mockery, experts in online threats say it’s foolish to think you’re too savvy to never fall for a professional scammer. “The reality is, criminals perpetuating fraud—whether via phone, email, or social media—are very good at social engineering,” says Selena Larson, a senior threat intelligence analyst at security firm Proofpoint, who describes Cowles as “extremely courageous.”

Manipulative tactics the scammers used against Cowles are common. They include, Larson says, “making someone afraid for themselves or their families, making them excited or enticed by the possibility of money or romance, or any number of heightened emotions to push them into making decisions they otherwise wouldn’t.” To protect yourself from scams like the one that hooked Cowles, Larson suggests being on high alert for anyone trying to isolate you from people in your life, and don’t trust someone posing as a government employee or celebrity. “Forcing a sense of urgency,” like asking for money immediately, is also a huge red flag. “If people are worried they are being targeted by fraudsters,” Larson says, “they should immediately break off contact and report the activity.”

Or you can adopt Cowles' new tactic: Never answer the phone.

Generative AI tools like ChatGPT are all the rage—including among hackers working on behalf of Russia, China, and North Korea, according to research published this week by Microsoft and OpenAI. While researchers note that they have “not identified significant attacks” that use large language models like those powering OpenAI’s ChatGPT, they did find widespread use of generative AI tools for research, reconnaissance, “basic scripting tasks,” and ways to improve code used to carry out cyberattacks. “Microsoft and OpenAI have not yet observed particularly novel or unique AI-enabled attack or abuse techniques resulting from threat actors’ usage of AI,” Microsoft wrote in a blog post outlining the research. “However, Microsoft and our partners continue to study this landscape closely.”

The US Department of Justice announced this week that it had disrupted a botnet controlled by APT28, a hacking group known as Fancy Bear that operates under Russia’s GRU military intelligence service. According to the DOJ, the hackers infected hundreds of routers used by homes and businesses with the “Moobot” malware, which the DOJ says is linked to a cybercriminal group. Fancy Bear hackers then used to Moobot to “install their own bespoke scripts and files that repurposed the botnet, turning it into a global cyber espionage platform,” according to the DOJ. To seize control of the botnet, the US government also used the Moobot malware to delete “stolen and malicious data” in the routers and then tweaked the routers’ firewalls to prevent the hackers from accessing them remotely. US attorney general Merrick Garland praised the operation in a statement as a successful effort to “dismantle the Russian government’s malicious cyber tools that endanger the security of the United States and our allies.”

Ransomware attacks frequently target hospitals, but few have had as widespread an impact as a strike against Romania’s health care system this week. Approximately 100 hospitals took their systems offline after attackers hit a popular hospital management system. Romanian officials say 25 hospitals had their data encrypted by the ransomware, which targeted the Hipocrate Information System (HIS) on the night of February 11. Another 75 hospitals voluntarily took their systems offline to avoid possible infection. The disruption has forced the hospitals to revert to paper records. The attackers, who have not yet been identified, demanded a ransom of 3.5 bitcoin, or around $180,000, to decrypt the files.

How to Not Get Scammed Out of $50,000

The prototype satellites hitched a ride on a Falcon 9 rocket.

Two prototype satellites for the Missile Defense Agency and four missile-tracking satellites for the US Space Force rode a SpaceX Falcon 9 rocket into orbit Wednesday from Florida's Space Coast.

These satellites are part of a new generation of spacecraft designed to track hypersonic missiles launched by China or Russia and perhaps emerging missile threats from Iran or North Korea, which are developing their own hypersonic weapons.

Hypersonic missiles are smaller and more maneuverable than conventional ballistic missiles, which the US military's legacy missile defense satellites can detect when they launch. Infrared sensors on the military's older-generation missile tracking satellites are tuned to pick out bright thermal signatures from missile exhaust.

The New Threat Paradigm

Hypersonic missiles represent a new challenge for the Space Force and the Missile Defense Agency (MDA). For one thing, ballistic missiles follow a predictable parabolic trajectory that takes them into space. Hypersonic missiles are smaller and comparatively dim, and they spend more time flying in Earth's atmosphere. Their maneuverability makes them difficult to track.

A nearly five-year-old military organization called the Space Development Agency (SDA) has launched 27 prototype satellites over the last year to prove the Pentagon's concept for a constellation of hundreds of small, relatively low-cost spacecraft in low-Earth orbit. This new fleet of satellites, which the SDA calls the Proliferated Warfighter Space Architecture, will eventually number hundreds of spacecraft to track missiles and relay data about their flight paths down to the ground. The tracking data will provide an early warning to those targeted by hypersonic missiles and help generate a firing solution for interceptors to shoot them down.

The SDA constellation combines conventional tactical radio links, laser inter-satellite communications, and wide-view infrared sensors. The agency, now part of the Space Force, plans to launch successive generations, or tranches, of small satellites, each introducing new technology. The SDA's approach relies on commercially available spacecraft and sensor technology and will be more resilient to attack from an adversary than the military's conventional space assets. Those legacy military satellites often cost hundreds of millions or billions of dollars apiece, with architectures that rely on small numbers of large satellites that might appear like a sitting duck to an adversary determined to inflict damage.

Four of the small SDA satellites and two larger spacecraft for the Missile Defense Agency were aboard a SpaceX Falcon 9 rocket when it lifted off from Cape Canaveral Space Force Station at 5:30 pm EST (2230 UTC) Wednesday.

The rocket headed northeast from Cape Canaveral to place the six payloads into low-Earth orbit. Officials from the Space Force declared the launch a success later Wednesday evening.

The SDA's four tracking satellites, built by L3Harris, are the last spacecraft the agency will launch in its prototype constellation, called Tranche 0. Beginning later this year, the SDA plans to kick off a rapid-fire launch campaign with SpaceX and United Launch Alliance to quickly build out its operational Tranche 1 constellation, with launches set to occur at one-month intervals to deploy approximately 150 satellites. Then, there will be a Tranche 2 constellation with more advanced sensor technologies.

The primary payloads aboard Wednesday's launch were for the Missile Defense Agency. These two Hypersonic and Ballistic Tracking Space Sensor (HBTSS) satellites, one supplied by L3Harris and the other by Northrop Grumman, will demonstrate medium field-of-view sensors. Those sensors can't cover as much territory as the SDA satellites but will provide more sensitive and detailed missile tracking data.

“Our advanced satellites on orbit will bring the integrated and resilient missile warning and defense capabilities the US requires against adversaries developing more advanced maneuverable missiles,” said Christopher Kubasik, chairman and CEO of L3Harris. “L3Harris delivered this advanced missile tracking capability on behalf of MDA and SDA on orbit in just over three years after work was authorized to proceed. We are proud to be a critical part of the new space sensing architecture.”

The HBTSS satellites, valued at more than $300 million, and the SDA's tracking prototypes will participate in joint military exercises in the coming months, where the wide-view SDA satellites will provide "cueing data" to the MDA's HBTSS spacecraft. The narrower field of view of the HBTSS satellites can provide more specific, target-quality data to a ground-based interceptor, according to a report last year published by the Congressional Research Service. Future tranches, or generations, of SDA satellites will incorporate the medium field-of-view sensing capability flying on the MDA's HBTSS satellites.

With SDA taking over the responsibility for making this technology operational, that will leave the MDA, which has historically flown its own missile-tracking satellites, focused on next-generation sensor development, an MDA spokesperson told Ars.

Military officials decided only last year to place the four SDA satellites on the same launch as the MDA's HBTSS mission. With all six satellites flying in the same orbital plane, there will be opportunities to see the same targets with both types of spacecraft and sensors. These targets may include scheduled US military missile tests or foreign launches.

"The intent is to be able to work with cooperative and noncooperative targets to be able to do our demonstrations," a senior SDA official said during a background briefing.

_This story originally appeared on_ _Ars Technica._

SpaceX Launched Military Satellites Designed to Track Hypersonic Missiles

A surprise disclosure of a national security threat by the House Intelligence chair was part of an effort to block legislation that aimed to limit cops and spies from buying Americans' private data.

The latest botched effort at salvaging a controversial US surveillance program collapsed this week thanks to a sabotage campaign by the United States House Intelligence Committee (HPSCI), crushing any hope of unraveling the program’s fate before Congress pivots to prevent a government shutdown in March.

An agreement struck between rival House committees fell apart on Wednesday after one side of the dispute—represented by HPSCI—ghosted fellow colleagues at a crucial hearing while working to poison a predetermined plan to usher a “compromise bill” to the floor.

A civil war between the House Judiciary and Intelligence Committees has crippled months of efforts to reauthorize Section 702 of the Foreign Intelligence Surveillance Act (FISA), an unpopular but crucial spy power, stunning the intelligence system and forcing security hawks to publicly argue in favor of surveillance tactics that even top spies acknowledge has been prone to abuse.

Witnesses to the events this week that forced House speaker Mike Johnson to shelve the latest Section 702 bill—the third of its kind to fail in as many months—say the leaders of HPSCI abandoned a deal that had been agreed to in private after weeks of negotiation. Sources familiar with the negotiations asked not to be identified, as none of them are authorized to speak publicly.

The impetus for killing the deal, WIRED has learned, was an amendment that would end the government’s ability to pay US companies for information rather than serving them with a warrant. This includes location data collected from cell phones that are capable in many cases of tracking people’s physical whereabouts almost constantly. The data is purportedly gathered for advertising purposes but is collected by data brokers and frequently sold to US spies and police agencies instead.

Senior aides say the HPSCI chair, Mike Turner, personally exploded the deal while refusing to appear for a hearing on Wednesday in which lawmakers were meant to decide the rules surrounding the vote. A congressional website shows that HPSCI staff had not filed one of the amendments meant to be discussed before the Rules Committee, suggesting that at no point in the day did Turner plan to attend.

Two senior sources on the Hill who are working for members with direct knowledge of the events but are not affiliated with either of the relevant committees say that while lawmakers waited on Turner to appear, he was meeting privately with Johnson and threatening to kill the bill he'd already signed off on.

At the same time, Turner and other HPSCI members were engaged in a floundering but possibly effective scheme to whip votes against any potential privacy enhancements, floating vague claims about an “urgent” threat against the US. Turner’s warning was later reported to concern Russia developing the capability to deploy nuclear weapons in space.

In a letter on Friday, four advocacy organizations signed a letter calling for Turner to step down from his role as Intel chair over what they called a “near-panic” started purely for “political gain.” The letter, first reported by Politico, was signed by members of groups FreedomWorks, Due Process Institute, Demand Progress, and Restore the Fourth.

Neither Johnson nor Turner’s offices responded to a request for comment.

Internal emails obtained by WIRED show that HPSCI’s ranking Democrat, Jim Himes, was likewise a part of that effort, his signature attached to a letter urging Democrats to schedule time to review “certain intelligence” in a HPSCI-operated “SCIF”—a room designed to securely host classified information. The letter implies an “urgent” threat involving foreign military capabilities that requires wide dissemination, all but ensuring it leaked to the press.

In news reports about that intelligence the following day, anonymous officials ensured it was stressed to reporters that Section 702 was essential to identifying the threat. More recent statements made by the White House and Senate intelligence staff that relay the risks the disclosure poses to classified “sources and methods” have generated skepticism around 702’s role.

Four senior aides tell WIRED that, despite speculation in _The New York Times_ that a foreign aid package related to Ukraine had been the impetus behind the effort to center the threat, Turner and Himes were motivated to instill paranoia in members that would inevitably raise doubts as to whether popular private reforms were simply too great a risk—namely those requiring the government to obtain warrants before accessing the private calls, texts, and emails of US persons, as well as “commercially available data” that can be used to monitor their whereabouts, both historically and in real-time.

This campaign—mirrored with less success by the FBI—appeared to backfire in a matter of hours, with Senate intelligence staff refusing to support Turner’s claims and instead issuing a statement that implied his disclosure had put classified sources at risk. The White House similarly dismissed the urgency communicated by Turner, warning that information related to the “threat” might expose intelligence sources.

Some aides critical of the tactics say nevertheless that they believe it may have been effective, shaking some members’ willingness to openly support reform.

“These tricks aren’t new,” said one aide. “They’re recycled, but increasingly transparent.” (“Every spymaster in history has exploited the mixed feeling of awe, gratitude, and fear that power-holders respond to in dealing with intelligence,” civil liberties attorney Frank Donner once said. It was 1971.)

By Wednesday afternoon, Himes, whose signature appeared on the initial “dear colleague” email announcing that the classified matter “should be known by all,” appeared to be seeking distance from the controversy, issuing a statement acknowledging the intel was “not a cause for panic” and saying the protection of sources is “a legal and sacred duty.”

After Johnson reluctantly killed the plans to move forward with a vote—and with it, any hope of resolving the battle over Section 702—intelligence officials began the task of redirecting the blame. The complexities of the issue, and general lack of knowledge among the public and press, aided significantly in that goal.

A Fox News report published Thursday morning, while accurately noting that it was Turner's threat that forced Johnson to cancel the vote, goes on to cite “sources close to the Intelligence Committee” who offered analysis of the events. The sources claimed that Turner was compelled to abandon the deal because the “compromise bill” had been sneakily altered in a manner that “totally screws FISA in terms of its ability to be a national security tool.”

While redirecting blame away from Turner and his cohorts, the claim is both false and deceptive, relying on assertions that, while farcical perhaps to legal experts, would be impossible for the public at large (and most of the press) to parse alone.

The text that Fox News’ intelligence sources are referring to—which can be read on the final page of the bill online—does nothing, in reality. It does not require or prevent anyone in the government from taking any action whatsoever. It also has no impact on FISA, the statute from which Section 702 derives its power.

The controversial text states that the nation’s top intelligence official “may submit” information to Congress regarding how “law enforcement agencies and the intelligence community” purchase “commercially available data about United States persons.” Essentially, it grants the intelligence community permission to do something that it does not actually need permission to do.

The language was included not to “totally screw” FISA, but to ensure that the phrase “commercially available data” appears at least once in the text, for reasons that are as benign as they are elusive to casual followers of legislative procedures.

One of the most popular amendments suggested to the Section 702 bill, discussed openly by lawmakers for months, is one that would prevent the government from purchasing data that normally requires a warrant. To counter arguments that these purchases are unrelated (which is to say, not “germane”) to the 702 program, the language in the final section, accomplishing nothing else, was added. A placeholder, effectively.

A senior source close to the Judiciary Committee said that it would have been impossible for Turner not to know the amendment was coming, and that the surprise expressed by his staff in the Fox News piece and elsewhere appeared to those in the know as pure theater.

Four aides, recordings of several public hearings, and a slew of reporting confirm that Turner had been aware for weeks, if not months, that restrictions on commercially available data would be one of the key amendments offered up by Judiciary members. The aides added that he'd also privately agreed to allow Judiciary members to offer their amendments. Prior to the bill being pulled, Representatives Warren Davidson and Zoe Lofgren had issued a relevant joint statement publicly: “It makes little sense to rein in warrantless surveillance under one authority when the government can simply fall back on other available techniques to acquire similar information,” they said.

Only after forcing Johnson to cancel the vote did the germaneness of the measure become a justification for tanking the entire process.

“No one actually thinks the Intelligence Committee cares about this,” says an aide working for a Judiciary member. “It’s the amendment they’re freaking out about. They don’t want the intelligence community to have to ask judges before they do anything.”

“For all the downplaying the agencies have done, telling us repeatedly they aren’t purchasing our data that often, Turner just blew weeks of negotiation to defend this one thing,” says the same aide. “To me, that says something about how much the government actually cares about this.”

_Update: 2/16/24, 3:35 pm ET: Added details about a letter calling for Turner to step down as Intel chair._

Leak of Russian ‘Threat’ Part of a Bid to Kill US Surveillance Reform, Sources Say

A new report cited 28 “verified” accounts on X that appear to be tied to sanctioned groups or individuals.

A watchdog group's investigation found that terrorist group Hezbollah and other US-sanctioned entities have accounts with paid check marks on X, the Elon Musk–owned social network that still resides at the Twitter.com domain.

The Tech Transparency Project (TTP), a nonprofit that is critical of Big Tech companies, said in a report on Wednesday that "X, the platform formerly known as Twitter, is providing premium, paid services to accounts for two leaders of a US-designated terrorist group and several other organizations sanctioned by the US government."

After buying Twitter for $44 billion, Musk started charging users for check marks that were previously intended to verify that an account was notable and authentic. "Along with the check marks, which are intended to confer legitimacy, X promises various perks for premium accounts, including the ability to post longer text and videos and greater visibility for some posts," the Tech Transparency Project report noted.

The Tech Transparency Project suggests that X may be violating US sanctions. "The accounts identified by TTP include two that apparently belong to the top leaders of Lebanon-based Hezbollah and others belonging to Iranian and Russian state-run media," the report said. "The fact that X requires users to pay a monthly or annual fee for premium service suggests that X is engaging in financial transactions with these accounts, a potential violation of US sanctions."

Some of the accounts were verified before Musk bought Twitter, but verification was a free service at the time. Musk's decision to charge for check marks means that X is "providing a premium, paid service to sanctioned entities," which may raise "new legal issues," the Tech Transparency Project said.

Report Details 28 Check-Marked Accounts

Musk's X charges $1,000 a month for a Verified Organizations subscription and last month added a basic tier for $200 a month. For individuals, the X Premium tiers that come with check marks cost $8 or $16 a month.

It's possible for US companies to receive a license from the government to engage in certain transactions with sanctioned entities, but it doesn't seem likely that X has such a license. X's rules explicitly prohibit users from purchasing X Premium "if you are a person with whom X is not permitted to have dealings under US and any other applicable economic sanctions and trade compliance law."

In all, the Tech Transparency Project said it found 28 "verified" accounts tied to sanctioned individuals or entities. These include individuals and groups listed by the US Treasury Department’s Office of Foreign Assets Control (OFAC) as Specially Designated Nationals.

"Of the 28 X accounts identified by TTP, 18 show they got verified after April 1, 2023, when X began requiring accounts to subscribe to paid plans to get a check mark. The other 10 were legacy verified accounts, which are required to pay for a subscription to retain their check marks," the group wrote, adding that it "found advertising in the replies to posts in 19 of the 28 accounts."

X issued the following statement on Wednesday: "X has a robust and secure approach in place for our monetization features, adhering to legal obligations, along with independent screening by our payments providers. Several of the accounts listed in the Tech Transparency Report are not directly named on sanction lists, while some others may have visible account check marks without receiving any services that would be subject to sanctions. Our teams have reviewed the report and will take action if necessary. We're always committed to ensuring that we maintain a safe, secure and compliant platform."

X Removes Some Check Marks

An account with the handle @SH\_NasrallahEng appears to be tied to Hezbollah leader Hassan Nasrallah, the TTP report said. The account had a check mark when we first checked it earlier Wednesday, but it has since been removed.

"The account, which has 93,600 followers, posts English-language Hezbollah messages and memes disparaging Israel and the US. It was created in October 2021 and verified in November 2023, the same month that Nasrallah threatened further escalation of Israel's war with Hamas," the report said.

Elon Musk’s X Gave Check Marks to Terrorist Group Leaders, Report Says

Prominent advocates for the rights of pregnant people are urging members of Congress to support legislation that would ban warrantless access to sensitive data as the White House fights against it.

One of the foremost women’s advocacy groups in the United States is urging members of Congress to support a ban on government agencies using private data brokers to acquire access to Americans’ private data without a warrant, a response that came late Tuesday following explosive revelations about the surveillance of women’s health clinics across the US.

The National Partnership for Women & Families, a nonpartisan nonprofit based in Washington, DC, is pressing federal lawmakers to throw support behind an amendment this week aimed at closing off a legal loophole through which police and spy agencies routinely circumvent the courts, buying information traditionally shielded by the country’s constitution.

The organization pointed to news in Politico on Tuesday which revealed a data broker tracked people’s visits to nearly 600 reproductive health clinics in the US and put that information up for sale. Politico cites a letter from US senator Ron Wyden to the Securities and Exchange Commission calling for an investigation into the company, Near Intelligence.

“This is not only a blatant violation of privacy but a cruel attempt to force pregnant people into reproductive health decisions that they should be making for themselves,” the National Partnership told members, according to an email obtained by WIRED. “Congress needs to curtail warrantless law enforcement access to people’s data in order to prevent government monitoring of abortion and pregnancy outcomes.”

The National Partnership for Women & Families has earned praise previously from former first ladies Michelle Obama and Hillary Clinton (during the latter’s tenure as secretary of state).

The data-broker amendment aimed at preventing government agencies from circumventing warrant requirements with money has gained widespread support in Congress. Likewise, recent polling shows up to 80 percent of Americans would support its passage. It faces powerful opposition, however, from the White House, the US military, and the nation’s spy agencies, which have acknowledged purchasing domestic internet data in recent years.

In December, pro-privacy members of the House Judiciary Committee attached language that would ban government data-broker purchases of information safeguarded under the nation’s Fourth Amendment to a House bill reauthorizing Section 702, a classified surveillance program designed to target foreigners on foreign soil that also routinely sweeps up the private communications of Americans: emails, calls, and text messages in their entirety.

Roughly a week later, pro-surveillance lawmakers on the House Intelligence Committee introduced their own bill reauthorizing the 702 program, which excluded the data-broker ban as well as other reforms included in the Judiciary Committee bill. One such reform, supported by the Judiciary Committee, would force federal analysts and investigators to get a warrant before examining the communications of Americans unavoidably intercepted under the sprawling 702 program.

US House leaders had tentatively agreed to a deal that would allow the entire House to vote on both bills, passing whichever received the most support. That plan fell through, however, and before the new year, lawmakers voted instead to extend the 702 program until mid-April. An effort to proactively block the Biden administration from taking advantage of this temporary extension, circumventing Congress and getting the program reauthorized for an additional year on its own, was ultimately rejected by House speaker Mike Johnson.

As a result, most House members remain confused as to when 702 surveillance would actually end if Congress fails to take action. Reformers say fomenting a sense of urgency to salvage the spy program—ultimately deemed vital even by many of its loudest critics—mostly plays into the administration’s hand, as it serves up “what-if” scenarios concerning possible terrorist attacks to lawmakers still on the fence. A group of senior congressional aides told WIRED last month that discussions about the program have been plagued for weeks by “scare tactics” and disinformation campaigns, with intelligence officials privately using images of Hamas to imply a growing domestic threat.

Rumors have circulated about a “secret session” being called this week, a rare procedure in which Congress meets behind closed doors. The session has been reported as called off, but a source with knowledge of recent developments tells WIRED that White House national security advisers are still expected to meet privately with lawmakers—one final attempt to dissuade them from supporting privacy reforms.

Last week, House speaker Mike Johnson and House minority leader Steve Scalise privately signed off on what they’d planned to advertise as a “compromise” bill, the latest in a string of schemes aimed at preserving the 702 program with as few changes as possible. It drew immediate criticism from civil liberties organizations such as the Brennan Center for Justice, which said it had been “carefully crafted” to preserve the “status quo.” The Electronic Privacy Information Center (EPIC) said the House leadership bill was a “compromise” in name only, aligning clearly with the priorities of the spy agencies over those fighting for reform.

Multiple sources, however, say the bill ultimately gained acceptance on the condition that members of both the House Judiciary and House Intelligence Committees would be allowed to offer amendments this week that would be subject to a floor vote. The amendment supported by the National Partnership for Women & Families is destined to be among them.

Police and intelligence agencies regularly purchase millions of dollars worth of sensitive information from data brokers each year, according to a December 2021 study of public records by the Center for Democracy & Technology (CDT), a civil-liberties-focused nonprofit. This data can include phone location data and health data collected by medical apps, which could be used to identify people seeking abortion care.

The Congressional Research Service (CRS), which provides Congress with legal and policy analysis, noted in 2022 that federal law includes “relatively little constraints” on law enforcement gaining access to sensitive data, including geolocation data and health data collected by apps and fitness trackers. The lack of constraints is particularly true for information sold by data brokers, which are “generally not regulated by any specific privacy statute,” according to CRS. While abortion-related information obtained from data brokers is known to have been used by anti-abortion activists, the CRS notes that it could equally be used by police investigating violations of state-level abortion laws.

The primary federal law regulating data broker activities is the FTC Act, which gives the US Federal Trade Commission the authority to penalize companies that fail to disclose how the data they sell may be used. In January, the FTC banned X-Mode Social, a Virginia-based data broker now named Outlogic, from selling “sensitive location data” that “could be used to track people’s visits to sensitive locations such as medical and reproductive health clinics, places of religious worship, and domestic abuse shelters” after the company allegedly failed to institute “appropriate safeguards” against the use of precise location data by third parties.

In July 2022, US president Joe Biden issued an executive order instructing the FTC chair to “consider actions” that aim to further “protect consumers’ privacy when seeking information about and provision of reproductive healthcare services.” The House Judiciary Committee’s amendment, which US spy agencies oppose, would strengthen these protective efforts far beyond the remit of the FTC Act.

In a “dear colleagues” email obtained by WIRED, Jerrold Nadler, the ranking Democrat on the Judiciary Committee, and Representative Zoe Lofgren wrote Wednesday that the so-called compromise bill “closely tracks” with the demands of the intelligence community, “bypassing commonsense reforms,” including the amendment now endorsed by the National Partnership for Women & Families, which Lofgren and Nadler describe as strictly written to stop the government from “buying its way around the Fourth Amendment.”

“The implications for Americans’ privacy rights are staggering,” they said, refuting claims that the data-broker issue is unrelated to surveillance conducted under the 702 program. “It makes little sense to rein in warrantless surveillance under one authority when the government can simply fall back on other available techniques to acquire similar information.”

Section 702 Surveillance Fight Pits the White House Opposite Reproductive Rights

Romantic chatbots collect huge amounts of data, provide vague information about how they use it, use weak password protections, and aren’t transparent, new research from Mozilla says.

You shouldn’t trust any answers a chatbot sends you. And you probably shouldn’t trust it with your personal information either. That’s especially true for “AI girlfriends” or “AI boyfriends,” according to new research.

An analysis into 11 so-called romance and companion chatbots, published on Wednesday by the Mozilla Foundation, has found a litany of security and privacy concerns with the bots. Collectively, the apps, which have been downloaded more than 100 million times on Android devices, gather huge amounts of people’s data; use trackers that send information to Google, Facebook, and companies in Russia and China; allow users to use weak passwords; and lack transparency about their ownership and the AI models that power them.

Since OpenAI unleashed ChatGPT on the world in November 2022, developers have raced to deploy large language models and create chatbots that people can interact with and pay to subscribe to. The Mozilla research provides a glimpse into how this gold rush may have neglected people’s privacy, and into tensions between emerging technologies and how they gather and use data. It also indicates how people’s chat messages could be abused by hackers.

Many “AI girlfriend” or romantic chatbot services look similar. They often feature AI-generated images of women which can be sexualized or sit alongside provocative messages. Mozilla’s researchers looked at a variety of chatbots including large and small apps, some of which purport to be “girlfriends.” Others offer people support through friendship or intimacy, or allow role-playing and other fantasies.

“These apps are designed to collect a ton of personal information,” says Jen Caltrider, the project lead for Mozilla’s Privacy Not Included team, which conducted the analysis. “They push you toward role-playing, a lot of sex, a lot of intimacy, a lot of sharing.” For instance, screenshots from the EVA AI chatbot show text saying “I love it when you send me your photos and voice,” and asking whether someone is “ready to share all your secrets and desires.”

Caltrider says there are multiple issues with these apps and websites. Many of the apps may not be clear about what data they are sharing with third parties, where they are based, or who creates them, Caltrider says, adding that some allow people to create weak passwords, while others provide little information about the AI they use. The apps analyzed all had different use cases and weaknesses.

Take Romantic AI, a service that allows you to “create your own AI girlfriend.” Promotional images on its homepage depict a chatbot sending a message saying,“Just bought new lingerie. Wanna see it?” The app’s privacy documents, according to the Mozilla analysis, say it won’t sell people’s data. However, when the researchers tested the app, they found it “sent out 24,354 ad trackers within one minute of use.” Romantic AI, like most of the companies highlighted in Mozilla’s research, did not respond to WIRED’s request for comment. Other apps monitored had hundreds of trackers.

In general, Caltrider says, the apps are not clear about what data they may share or sell, or exactly how they use some of that information. “The legal documentation was vague, hard to understand, not very specific—kind of boilerplate stuff,” Caltrider says, adding that this may reduce the trust people should have in the companies.

Source

Wired