The easy access that scammers have to sophisticated AI tools means everything from emails to video calls can’t be trusted.

Imagine you meet someone new. Be it on a dating app or social media, you chance across each other online and get to talking. They’re genuine and relatable, so you quickly take it out of the DMs to a platform like Telegram or WhatsApp. You exchange photos and even video call each over. You start to get comfortable. Then, suddenly, they bring up money.

They need you to cover the cost of their Wi-Fi access, maybe. Or they’re trying out this new cryptocurrency. You should really get in on it early! And then, only after it’s too late, you realize that the person you were talking to was in fact not real at all.

They were a real-time AI-generated deepfake hiding the face of someone running a scam.

This scenario might sound too dystopian or science-fictional to be true, but it has happened to countless people already. With the spike in the capabilities of generative AI over the past few years, scammers can now create realistic fake faces and voices to mask their own in real time. And experts warn that those deepfakes can supercharge a dizzying variety of online scams, from romance to employment to tax fraud.

David Maimon, the head of fraud insights at identity verification firm SentiLink and a professor of criminology at Georgia State University, has been tracking the evolution of AI romance scams and other kinds of AI fraud for the past six years. “We’re seeing a dramatic increase in the volume of deepfakes, especially in comparison to 2023 and 2024,” Maimon says.

“It wasn’t a whole lot. We’re talking about maybe four or five a month,” he says. “Now, we’re seeing hundreds of these on a monthly basis across the board, which is mind-boggling.”

Deepfakes are already being used in a variety of online scams. One finance worker in Hong Kong, for example, paid $25 million to a scammer posing as the company’s chief financial officer in a deepfaked video call. Some deepfake scammers have even posted instructional videos on YouTube, which have a disclaimer as being for “pranks and educational purposes only.” Those videos usually open with a romance scam call, where an AI-generated handsome young man is talking to an older woman.

More traditional deepfakes—such as a pre-rendered video of a celebrity or politician, rather than a live fake—have also become more prevalent. Last year, a retiree in New Zealand lost around $133,000 to a cryptocurrency investment scam after seeing a Facebook advertisement featuring a deepfake of the country’s prime minister encouraging people to buy in.

Maimon says SentiLink has started to see deepfakes used to create bank accounts in order to lease an apartment or engage in tax refund fraud. He says an increasing number of companies have also seen deepfakes in video job interviews.

“ Anything that requires folks to be online and which supports the opportunity of swapping faces with someone—that will be available and open for fraud to take advantage of,” Maimon says.

Part of the reason for this increase is that the barriers for creating deepfakes are getting lower. There are a lot of easily accessible AI tools that can generate realistic faces and a lot of tools that can animate those faces or create full-length videos out of them. Scammers often use images and videos of real people, deepfaked to slightly change their faces or alter what they’re saying, to target their loved ones or hijack their public influence.

Matt Groh, a professor of management at Northwestern University who researches people’s ability to detect deepfakes, says that point-and-click generative AI tools make it much easier to make small, believable changes to already-existing media.

“If there’s an image of you on the internet, that would be enough to manipulate a face to look like it’s saying something that you haven’t said before or doing something you haven’t done before,” Groh says.

It’s not just fake video that you need to be worried about. With a few clips of audio, it’s also possible to make a believable copy of somebody’s voice. One study in 2023 found that humans failed to detect deepfake audio over a quarter of the time.

“ Just a single image and five seconds of audio online mean that it’s definitely possible for a scammer to make some kind of realistic deepfake of you,” Groh says.

Deepfakes are becoming more pervasive in contexts other than outright scams. Social media has been flooded over the past year with AI-generated “influencers” stealing content from adult creators by deepfaking new faces onto their bodies and monetizing the resulting videos. Deepfakes have even bled over into geopolitics, like when the mayors of multiple European capital cities held video calls with a fake version of the mayor of Kyiv, Ukraine. People have started using deepfakes for personal reasons, like bringing back a dead relative or creating an avatar of a victim to testify in court.

So, if deepfakes are everywhere, how do you spot one? The answer is not technology. A number of technology companies, including OpenAI, have launched deepfake detection tools. Researchers have also proposed mechanisms to detect deepfakes based on things like light reflected in a person’s eyes or inconsistent facial movements, and have started investigating how to implement them in real time.

But those models often cannot reliably detect different kinds of AI fakes. OpenAI’s model, for example, is specifically designed only to report content generated with the company’s own Dall-E 3 tool but not other image generation models.

There’s also the risk that scammers can abuse AI detectors by repeatedly tweaking their content until it fools the software.

“ The major thing we have to understand is that the technology we have right now is not good enough to detect those deepfakes,” Maimon says. “ We’re still very much behind.”

For now, as video deepfakes get more popular, the best way to detect one relies on humans. Studies on deepfake detection show that people are best at distinguishing whether videos are real or fake, as opposed to just audio or text content, and are in some cases even better than leading detection models.

Groh’s team conducted one study which found that taking more time to determine whether an image was real or fake led to a significant increase in accuracy, by up to eight percentage points for just 10 seconds of viewing time.

“ This sounds almost so simple,” Groh says. “But if you spend just a couple extra seconds, that leads to way higher rates of being able to distinguish an image as real or fake. One of the ways for any regular person to just be a little bit less susceptible to a scam is to ask, ‘Does this look actually real?’ And if you just do that for a few extra seconds, we’re all going to be a little bit better off.”

Deepfakes’ popularity could be a double-edged sword for scammers, Groh says. The more widespread they are, the more people will be familiar with them and know what to look for.

That familiarity has paid off in some cases. Last summer, a Ferrari executive received a call from someone claiming to be the CEO of the company. The person convincingly emulated the CEO’s voice but abruptly hung up the call when the executive tried to verify their identity by asking what book the CEO had recommended just days earlier. The CEO of WPP, the world’s biggest advertising agency, was also unsuccessfully targeted by a similar deepfake scam.

“I think there’s a balancing act going on,” Groh says. “ We definitely have technology today that is generally hard for people to identify. But at the same time, once you know that there’s a point-and-click tool that allows you to transform one element into something else, everyone becomes a lot more skeptical.”

Deepfake Scams Are Distorting Reality Itself

A quantum computer will likely one day be able to break the encryption protecting the world's secrets. See how much faster such a machine could decrypt a password compared to a present-day supercomputer.

WIRED is where tomorrow is realized. It is the essential source of information and ideas that make sense of a world in constant transformation. The WIRED conversation illuminates how technology is changing every aspect of our lives—from culture to business, science to design. The breakthroughs and innovations that we uncover lead to new ways of thinking, new connections, and new industries.

See How Much Faster a Quantum Computer Will Crack Encryption

Seems bad out there. Unfortunately, it can always get worse. From evil hacker AI to world-changing cyberattacks, WIRED envisions the future you haven't prepared for.

You're Not Ready

Everyone knows what it’s like to lose cell service. A burgeoning open source project called Meshtastic is filling the gap for when you’re in the middle of nowhere—or when disaster strikes.

Hypothetical: You wake up tomorrow morning to find a superstorm that developed overnight thanks to climate change has sparked a chain of events that abruptly ushers in a new ice age and alters human society as we know it. (Yes, this is the plot of _The Day After Tomorrow_. Stick with us.) All the communication networks you relied on are down. Your phone is basically worthless. The internet has functionally ceased to exist. But you need to connect with people you trust to get help and survive. What do you do? More importantly, how did you prepare?

Less Hollywood-esque versions of having no cell service or Wi-Fi happen all the time, of course; maybe you’re hiking in a secluded area, white-knuckling through a major natural disaster, or living under a repressive regime that cuts internet access to quash public protests. Fortunately, for all these scenarios, there’s a low-budget solution: Meshtastic.

Meshtastic is a program that enables devices to send text messages over long distances without needing Wi-Fi or cell service. Long range radio (LoRa) nodes help pass messages along, forming a network of devices that can talk to each other even in remote areas. Messages hop from device to device, with each node relaying messages it hasn't seen before—extending the network’s reach across miles using minimal power. That is to say, Meshtastic is designed specifically for sending text messages over free-to-use radio frequencies to both groups and individuals, even when cell service and internet connections are nowhere to be found.

“The cool thing about Meshtastic is that it’s like a radio infrastructure without the infrastructure. It’s ad hoc,” says Eric Kristoff, a volunteer member of the Chicago chapter of the Mars Society, a nonprofit that advocates for the human exploration and colonization of Mars.

Kristoff says the group has been testing the use of Meshtastic as a way to give Mars Society “analog astronauts” the ability to communicate and keep track of each others’ locations without the use of earthly infrastructure.

“We have a set of Meshtastic T-Echo radios, about the size of a deck of cards, and they are worn on the person of the analog astronaut,” he says.

The radios that use Meshtastic cost roughly $30 (though you can spend two, three, or four times that if you want to). And because they operate over unlicensed radio frequencies on a network created by personal devices, they’re essentially free to use. Each message is end-to-end encrypted, ensuring privacy while it’s relayed through the network. And Meshtastic’s optional location-tracking capabilities give people a way to monitor their communities and keep tabs on their kids—or fellow analog astronauts—without using invasive, data-hungry apps.

Kristoff says that Mars Society members will take weeks-long excursions in remote areas with little cellular or Wi-Fi connectivity, which creates additional risks.

“There is heat stroke. We are two hours from the nearest hospital. If you go too far from the campus, it can get dangerous,” Kristoff says of the experience. “So anytime there’s a risk, the risk is made worse if people don’t know where you are.”

Most Meshtatic devices currently on the market need to pair with a phone over Bluetooth to function as a texting alternative. Some devices are just a radio, antenna, and battery, with the expectation that you’ll make the housing yourself. The radio does all the device-to-device communication, while the iOS and Android apps or the web client let you read and compose messages that are received or sent over the network—no service plan needed.

The apps also allow you to see the approximate location of nearby nodes and a map of the Meshtastic network. But fancier stand-alone devices are already available, like a line of Meshtastic-enabled gadgets from maker-friendly tech firm LilyGo that, in addition to the T-Echo model used by Mars Society members, includes Blackberry-like handhelds with their own keyboards, a smartphone-like device with an e-paper screen, and even a Meshtastic-enabled smartwatch.

Meshtastic was created by technologist Kevin Hester in early 2020 as a way to communicate while doing “any hobby where you don’t have reliable internet access,” and it remains a grassroots endeavor, with established local communities spanning from Argentina to China that are ripe with a DIY ethos. The software itself is open source, meaning anyone can theoretically contribute, and hundreds have. Still, as with many open source projects, a core group of volunteer developers help maintain the Meshtastic firmware, mobile apps, and more.

Jonathan Bennett, a self-described “Linux guy” who upgraded Meshtastic to stronger end-to-end encryption for direct messaging and keeps the software working on Linux, says he first got involved in the project after a listener of one his podcasts wanted a way to communicate with friends while attending a festival where the cell network could get overloaded.

“I put my open source enthusiast hat on and I went looking, and I came across Meshtastic,” he says. “And it immediately tickled my interest.”

Bennett says he ultimately connected with Garth Vander Houwen, a C# developer who wrote Meshtastic’s iOS app, and Ben Meadors, another C# developer who took on maintaining the Android app, web client, firmware, and other parts of the Meshtastic ecosystem after Hester needed to step back due to health issues.

Like Bennett, Vander Houwen and Meadors got involved with Meshtastic while looking for solutions to real-life problems. Vander Houwen, an iPhone user, says he found Meshtastic while on the hunt for a way to communicate as he hiked on remote trails in the Seattle area, just to find that it only had an Android app. He decided to write the iOS app himself. “So the fact that there was not an iOS app for Meshtastic was kind of how I got started,” he says, “and it's been a lot of fun.”

Meadors says he came to Meshtastic after a dangerous tornado hit his home state of Arkansas, causing major damage. “My kind of initial use case was honestly a backup communication for storm-related outages,” Meadors says. After taking part in the cleanup effort around Little Rock, he realized the value of a decentralized, off-grid communication network like Meshtastic. “It's just really handy to have anywhere where you’ve got a limited connection to the grid.”

None of which is to say you should throw your cell phone in the sea and go all-in on Meshtastic. At least not yet. First, getting into the world of LoRa remains a little bit technical, so if the idea of “flashing” your device with new firmware makes you instinctively pick up your phone to scroll TikTok, it might not be the hobby for you. Even if you are tech savvy, the system has some notable limitations.

Using the decentralized mesh network requires having your Meshtastic device in range of at least one other radio; obstructions like buildings, trees, and hills or mountains can prevent the line-of-sight communication needed to join the mesh network. This means it may only be reliable when there’s a variety of other Meshtastic nodes in the area.

Next is what Meadors calls the “narrowness” of the network’s technical capabilities. “One of the most frequent things that we get is, ‘Can I replace the internet with this?’ No, no you cannot,” he says. “You can send text messages.” Mercifully, that does include emoji.

Photograph: Michael Tessier

This may be obvious, but you also need to have a network set up before disaster hits, Meadors says. So, set up anyone who you might need to communicate with during a cell and internet blackout before it actually happens. And, due to relatively frequent firmware updates, you can’t just toss your device in a bug-out bag and forget about it. But “if it's something that you actually use, like if you pull it out and use it once a month, you'll be good to go,” Bennett says.

Then there’s the issue of raw bandwidth. This limitation can cause issues when a lot of people are trying to use the network at the same time. At a ham radio convention in Dayton, Ohio, last year (yes, it’s called Hamvention), the Meshtastic network crashed after someone ran a program that flooded the network with additional traffic, pushing the Meshtastic network to its limits.

“Because literally one person turned on this MQTT bridge, which then joined the rest of us into this mesh in a metal building in Dayton, it crashed the whole mesh immediately,” Vander Houwen says.

After this incident, Vander Houwen, Bennett, and Meadors went to work to prepare for the upcoming Defcon hacker convention in Las Vegas, ultimately releasing a special firmware for the much larger event that Vander Houwen estimates allows “somewhere between 2,000 and 2,500 nodes” to operate on the network simultaneously. A similar firmware released ahead of the 2025 Hamvention in May drew praise from the community.

Despite Meshtastic’s limitations, its promise as a backup communication system—and the sheer fun you can have with it—continues to pull in new enthusiasts. The Android app alone has drawn thousands of reviews, and the Meshtastic subreddit has grown to nearly 50,000 members. Some municipalities are even hoping to launch Meshtastic networks to help protect their communities in the event of natural disasters.

For Bennett, Meadors, and Vander Houwen, they’re excited to not just see the number of Meshtastic nodes increase, but to see the technology develop into something anyone can use without having to become an enthusiast or “analog astronaut” at all.

“I think the biggest thing for me too is that it’s not just accessible from the aspect of the hardware being available to more people. I want to make the software more accessible,” Meadors says. “I want to make the experience such that I can hand this device to anybody and have them download the app and start messaging. We've come a long way. I think there's still some room to grow there.”

_Updated at 10:25 am ET, June 4, 2025: Corrected a misspelling of Ben Meadors' surname._

The Texting Network for the End of the World

GPS jamming and spoofing attacks are on the rise. If the global navigation system the US relies on were to go down entirely, it would send the world into unprecedented chaos.

Around 12,500 miles above our heads, the satellites that make up the Global Positioning System (GPS) quietly keep the world running. A blackout would result in almost instantaneous chaos.

“You would see traffic jams, a lot more traffic accidents, because transportation is going to see the first most immediate impact,” says Dana Goward, the founder of the Resilient Navigation and Timing Foundation, a charity which works to strengthen GPS.

Thousands of planes in the air, which use GPS among other systems for navigation and precision landing, would face a wave of uncertainty. Then other critical parts of society—from financial transactions to energy production systems—which have come to rely upon the precision positioning, navigation and timing (PNT) provided by the US-owned constellation of 31 GPS satellites may start to stutter. The ripples would be felt around the world.

“If it was a catastrophic moment that happened at a blink of an eye and we lost GPS entirely, you would see this global seizure of everything that moves, every piece of data that moves, every human that moves. All of that would shut down,” says Erik Daehler, the vice president of defense, satellites, and spacecraft systems at Sierra Space. The timing signals included in GPS would be one of the most impactful losses. Cell phone connections would likely collapse. Billions would quickly be wiped from stock markets amid the disruption.

A GPS outage could be particularly ruinous to the United States, which has a heavy reliance on its sovereign space system and has dragged its feet in building backups that can provide the required resilience needed to keep the country running. The US has fallen behind, the National Space-based PNT Advisory Board warned last year. In contrast, China has reinforced its own more modern satellite navigation system—BeiDou—with a sprawling network of fiber-optic cables and terrestrial radio signals.

The conditions needed to cause the entire GPS network to be entirely knocked out would be extraordinary and likely would come with wider societal ramifications. Such an outage, for instance, could be caused by China or Russia firing anti-satellite weapons against the GPS satellites (the US also has anti-satellite weapons), a powerful geomagnetic storm, or an escalation in the capabilities of electronic warfare.

Despite the improbability of a total outage, GPS isn’t infallible. It has its demons. “What really happens is, regionally, GPS gets messed with and jammed and interfered with on a regular basis,” Daehler says. Thousands of planes and ships are having their GPS interfered with each week, and signals are regularly disrupted around war zones.

“America is not well prepared at all,” Goward says. More should be done to build out PNT systems that can act as a backstop to the space-based GPS signals, he says. “There’s not a general overall awareness. We certainly don’t have a resilient PNT architecture or a PNT architecture of any kind other than GPS.”

The GPS constellation of 31 satellites, which has received several hardware updates over the years, has been in operation for the past 40 years. The system typically broadcasts at 100 percent availability and provides accurate location data to within 7 meters.

The GPS satellites are just one of the four so-called global navigation satellite systems (GNSS) in operation. As well as China’s BeiDou, there is Russia’s GLONASS and Europe’s Galileo constellation. Over the past half decade, though, GNSS signals have increasingly been attacked as the technology to disrupt them has become cheaper and more sophisticated. Most commonly, disruption happens around Russia, Israel, Myanmar, the South China sea, areas of the Middle East, and the Baltic countries in Europe.

Broadly, there are two main forms of attack against GNSS signals: jamming and spoofing. Jamming involves blocking signals so that positioning isn’t available, while spoofing involves creating mock signals that make something appear somewhere else on the map. Ships have been made to appear inland at airports, while planes are made to look like they are flying in tight circles. In one video shared by the Resilient Navigation and Timing Foundation that appears to show GPS interference, a plane’s systems blast out a warning message to “pull up” when its pilots reported they were flying higher than Mount Everest.

“I’m most concerned about aviation,” says Todd Humphreys, the director of the University of Texas at Austin’s radio navigation laboratory. “At least one fatal aviation accident in Europe can be traced to GNSS interference as a primary cause. A deliberate attack against US aviation, as opposed to the collateral attacks in Europe, would cause astounding economic harm.” The number of spoofing incidents last year was 500 percent higher than in 2023, according to aviation officials.

The US Space Force, which is responsible for the GPS satellites, did not respond to a request for comment for this article from WIRED.

Across the US, PNT data is crucial to almost all critical infrastructure—from communications and health care monitoring systems to food production and wastewater management—but GPS is often the “sole” source of this information, according to the Cybersecurity and Infrastructure Security Agency, making the systems more vulnerable. (The military uses a more robust GPS setup than commercial applications).

“There is no one sector that doesn’t use GPS, and some are more reliant than others. Users in these sectors are not all acutely aware of the risks associated with their dependency on it and the ways that the system can be disrupted or degraded,” says Caitlin Durkovich, a former national security official and critical infrastructure expert.

Building a “layered” approach could help to make GPS less vulnerable to attack, experts say. Both Europe’s Galileo and China’s BeiDou are newer than GPS and, in some ways, more resilient. Last year, the National Space–based PNT Advisory Board produced a comparison of GPS and BeiDou that flagged a broader series of backups to Beijing’s system.

While GPS satellites are located only in middle Earth orbit, BeiDou has satellites in multiple orbits and is further along in deploying them into low Earth orbit. China also has a terrestrial radio broadcast network, called eLoran, and has laid 20,000 kilometers of fiber-optic cables that link up with 295 timing centers to broadcast alternatives.

“In the case of BeiDou, the system’s enhanced resiliency and capability should be considered an element of ‘soft power’ and an element of great power competition,” the advisory board wrote last year. The board, led by Admiral Thad Allen, a former leader of the US Coast Guard, called for a more joined-up approach to managing PNT across the US government and for GPS to be specifically designated as “critical infrastructure.”

On April 26, 2024, the first of two Finnair flights were forced to turn around due to GPS interference likely carried out by Russia. After a second aircraft was diverted the following day, Finnair suspended its daily flights between Helsinki and Tartu, Estonia. _Source: AirNav_

“I think there has to be a federal role in this, both because the system and signals are operated and provisioned by the federal government. But because of the complexity of the system and the fact that you need a common standard,” Durkovich says.

“We’d like to see a core national PNT architecture,” Goward says. “Then we would suggest some form of fiber network and a terrestrial broadcast. We think it would be a substantial deterrent and it would actually make space-based systems safer because folks would be less likely to interfere with it.”

Across the country, there are various levels of backup systems in place that have been sporadically introduced and multiple ongoing efforts to improve the GPS setup. Financial institutions, for instance, have been deploying atomic clocks to ensure they have backups for the timing element provided by GPS and telecoms networks have some capacity in place.

“It’s not to say that the US doesn’t have a robust timing infrastructure, actually it’s quite robust,” says Jeremy Bennington, the vice president of PNT Assurance at Spirent Communications, adding that much of it is spread across commercial entities, a stark difference to China’s national approach. “I do think that a backup is going to be required so that you end up with that layered approach.”

The calls to modernize PNT have increasingly become more urgent. In 2020, a first-term Trump executive order called for making PNT systems more resilient. At the end of March this year, the Federal Communications Commission opened an inquiry to identify GPS alternatives that can provide backups. “Relying on GPS alone as the primary source of PNT data leaves America exposed to a single point of failure and leaves our PNT system open to disruption or manipulation by adversaries,” the FCC said at the time.

There are multiple ways to add more resilience and upgrade the existing GPS system. The military has long been working on upgrades to be used in defense situations. Bennington says that GPS satellites could be added to other orbits and the further rollout of more capable signals. Daehler and colleagues at Sierra Space are working on creating ways to reduce the impact of jamming and spoofing.

Lisa Dyer, the executive director of the GPS Innovation Alliance, says the GPS system could build in authentication to confirm its signals are genuine, like Galileo and BeiDou. Dyer says that rolling out the newer L5 signal can also build in more protection for planes and aviation. “To me that's an important national objective of the United States: that GPS remains the de facto international navigation standard,” Dyer says.

There are also hardware updates happening, though some of them are slow and have dragged on for years. The US Space Force has recently been funding multiple companies to develop low Earth orbit satellite GPS constellations and quickly launching systems into space. Elsewhere, quantum technologies are being used to create new navigation systems. SandboxAQ, a Google spinout, is working on magnetic navigation.

Ultimately, as well as better government management around GPS, organizations need to spend money to upgrade their systems and protections, Bennington says. That means spending money. “If GPS jamming or spoofing were to happen at any major airport, whether it's Heathrow, Frankfurt, Munich, New York, the amount of cancellation and delays in the cost incurred by the airlines just in several hours would be more than the cost to upgrade their fleets,” he says.

A GPS Blackout Would Shut Down the World

In the very near future, victory will belong to the savvy blackhat hacker who uses AI to generate code at scale.

In the near future one hacker may be able to unleash 20 zero-day attacks on different systems across the world all at once. Polymorphic malware could rampage across a codebase, using a bespoke generative AI system to rewrite itself as it learns and adapts. Armies of script kiddies could use purpose-built LLMs to unleash a torrent of malicious code at the push of a button.

Case in point: as of this writing, an AI system is sitting at the top of several leaderboards on HackerOne—an enterprise bug bounty system. The AI is XBOW, a system aimed at whitehat pentesters that “autonomously finds and exploits vulnerabilities in 75 percent of web benchmarks,” according to the company’s website.

AI-assisted hackers are a major fear in the cybersecurity industry, even if their potential hasn’t quite been realized yet. “I compare it to being on an emergency landing on an aircraft where it’s like ‘brace, brace, brace’ but we still have yet to impact anything,” Hayden Smith, the cofounder of security company Hunted Labs, tells WIRED. “We’re still waiting to have that mass event.”

Generative AI has made it easier for anyone to code. The LLMs improve every day, new models spit out more efficient code, and companies like Microsoft say they’re using AI agents to help write their codebase. Anyone can spit out a Python script using ChatGPT now, and vibe coding—asking an AI to write code for you, even if you don’t have much of an idea how to do it yourself—is popular; but there’s also vibe hacking.

“We’re going to see vibe hacking. And people without previous knowledge or deep knowledge will be able to tell AI what it wants to create and be able to go ahead and get that problem solved,” Katie Moussouris, the founder and CEO of Luta Security, tells WIRED.

Vibe hacking frontends have existed since 2023. Back then, a purpose-built LLM for generating malicious code called WormGPT spread on Discord groups, Telegram servers, and darknet forums. When security professionals and the media discovered it, its creators pulled the plug.

WormGPT faded away, but other services that billed themselves as blackhat LLMs, like FraudGPT, replaced it. But WormGPT’s successors had problems. As security firm Abnormal AI notes, many of these apps may have just been jailbroken versions of ChatGPT with some extra code to make them appear as if they were a stand-alone product.

Better then, if you’re a bad actor, to just go to the source. ChatGPT, Gemini, and Claude are easily jailbroken. Most LLMs have guard rails that prevent them from generating malicious code, but there are whole communities online dedicated to bypassing those guardrails. Anthropic even offers a bug bounty to people who discover new ones in Claude.

“It’s very important to us that we develop our models safely,” an OpenAI spokesperson tells WIRED. “We take steps to reduce the risk of malicious use, and we’re continually improving safeguards to make our models more robust against exploits like jailbreaks. For example, you can read our research and approach to jailbreaks in the GPT-4.5 system card, or in the OpenAI o3 and o4-mini system card.”

Google did not respond to a request for comment.

In 2023, security researchers at Trend Micro got ChatGPT to generate malicious code by prompting it into the role of a security researcher and pentester. ChatGPT would then happily generate PowerShell scripts based on databases of malicious code.

“You can use it to create malware,” Moussouris says. “The easiest way to get around those safeguards put in place by the makers of the AI models is to say that you’re competing in a capture-the-flag exercise, and it will happily generate malicious code for you.”

Unsophisticated actors like script kiddies are an age-old problem in the world of cybersecurity, and AI may well amplify their profile. “It lowers the barrier to entry to cybercrime,” Hayley Benedict, a Cyber Intelligence Analyst at RANE, tells WIRED.

But, she says, the real threat may come from established hacking groups who will use AI to further enhance their already fearsome abilities.

“It’s the hackers that already have the capabilities and already have these operations,” she says. “It’s being able to drastically scale up these cybercriminal operations, and they can create the malicious code a lot faster.”

Moussouris agrees. “The acceleration is what is going to make it extremely difficult to control,” she says.

Hunted Labs’ Smith also says that the real threat of AI-generated code is in the hands of someone who already knows the code in and out who uses it to scale up an attack. “When you’re working with someone who has deep experience and you combine that with, ‘Hey, I can do things a lot faster that otherwise would have taken me a couple days or three days, and now it takes me 30 minutes.’ That's a really interesting and dynamic part of the situation,” he says.

According to Smith, an experienced hacker could design a system that defeats multiple security protections and learns as it goes. The malicious bit of code would rewrite its malicious payload as it learns on the fly. “That would be completely insane and difficult to triage,” he says.

Smith imagines a world where 20 zero-day events all happen at the same time. “That makes it a little bit more scary,” he says.

Moussouris says that the tools to make that kind of attack a reality exist now. “They are good enough in the hands of a good enough operator,” she says, but AI is not quite good enough yet for an inexperienced hacker to operate hands-off.

“We’re not quite there in terms of AI being able to fully take over the function of a human in offensive security,” she says.

The primal fear that chatbot code sparks is that anyone will be able to do it, but the reality is that a sophisticated actor with deep knowledge of existing code is much more frightening. XBOW may be the closest thing to an autonomous “AI hacker” that exists in the wild, and it’s the creation of a team of more than 20 skilled people whose previous work experience includes GitHub, Microsoft, and a half a dozen assorted security companies.

It also points to another truth. “The best defense against a bad guy with AI is a good guy with AI,” Benedict says.

For Moussouris, the use of AI by both blackhats and whitehats is just the next evolution of a cybersecurity arms race she’s watched unfold over 30 years. “It went from: ‘I’m going to perform this hack manually or create my own custom exploit,’ to, ‘I’m going to create a tool that anyone can run and perform some of these checks automatically,’” she says.

“AI is just another tool in the toolbox, and those who do know how to steer it appropriately now are going to be the ones that make those vibey frontends that anyone could use.”

The Rise of ‘Vibe Hacking’ Is the Next AI Nightmare

A major cyberattack on the US electrical grid has long worried security experts. Such an attack wouldn’t be easy. But if an adversary pulled it off, it’d be lights out in more ways than one.

When the lights went out across the Iberian Peninsula in April, everything ground to a halt. Scores of people were trapped in Madrid’s underground metro system. Hospitals in Lisbon had to switch to emergency generators. Internet service as far away as Greenland and Morocco went down.

While the cause remains unclear, the actual damage to the Iberian power grid—and the people it serves—was relatively minor. Less than 24 hours after the outage began, the region’s electricity operators managed to get the grid back online.

Even if things could have been much worse, the outage was both an unnerving reminder of how suddenly things can go offline.

For years, cybersecurity professionals, watchdogs, and government agencies have warned that a malicious cyberattack on the US power grid could be devastating. With ample evidence that state-sponsored hacking groups are eyeing the decentralized and deeply vulnerable power grid, the risk is more acute than ever.

Case in point: Hackers, believed to be linked to the Chinese government, spent years exploiting vulnerabilities in critical infrastructure across the mainland United States and Guam to obtain access to their systems. The operations, dubbed Volt Typhoon, could have used this access to shut down or disconnect parts of the American power grid—throwing millions into the dark. The effort was, luckily, disrupted and the vulnerabilities patched. Still, it is an unnerving illustration of just how vulnerable the electric system truly is.

We know what such a hack could look like. In 2015, Ukraine experienced the world’s first large-scale cyberattack on an electrical grid. A Russian military intelligence unit known as Sandworm disconnected various substations from the central grid and knocked hundreds of thousands of people offline.

The attack on Ukraine was repaired quickly, but cybersecurity experts have been warning for years that the next one might be more devastating.

Unlike Ukraine, America does not have a single power grid—it has three large interconnections, broken down into a network of smaller regional systems, some of which stretch into Canada. Most of the East is on one grid, most of the West is on another, while Texas and Alaska run their own interconnections. Keeping these networks running is a wildly complicated effort: There are thousands of utility operations, tens of thousands of substations, and hundreds of thousands of miles of high-voltage transmission lines.

Photograph: Michael Tessier

To some degree, this decentralized network is an asset, as it means there is no core vulnerability that risks knocking the entire country offline. But the interconnections mean that a failure in one corner of the grid could cause a cascade that takes down the entire system.

In 2018, researchers from Northwestern University ran large-scale models, gaming out what would happen if parts of the grid failed. They found that, generally, the American power grid was resilient. However, they found that about 10 percent of power lines in the US were susceptible to the kind of failure that could trigger this domino effect under some conditions. A 2022 study that looked at possible disruptions to the Texas grid also found that, in some cases, a relatively small disruption could cause a series of downstream outages “rapidly in succession.”

This means that even if malicious actors manage to take only a small number of nodes in the network offline, it has the potential to do enormous downstream damage.

Insurance underwriter Lloyd’s of London has looked at the effects of such an outage. In this hypothetical, first drafted in 2015 but updated in the years since, Lloyd’s estimates that a Trojan virus that manages to infect just 50 generators—removing 10 percent of the grid’s total power—can trigger this cascade effect and knock out power for most of the East Coast, including New York City and Washington, DC. The Lloyd’s report states that this is an “extreme” but “not unrealistic scenario.”

“Images of a dark New York City make front pages worldwide,” they write, “accompanied by photographs of citizens stuck underground for hours on stranded subway cars and in elevators in the summer heat.”

These rolling blackouts would stretch through 36 states over the course of a day, throwing some 93 million people into the dark. It could take up to three days for half of those people to get back online—while hardware damage and other problems could require up to three weeks to fix.

As the outages continue, more difficulties arise. The analysts warn that an information campaign running parallel to the cyberattack could prompt strikes, protests, or general unrest.

In 2016, then Federal Emergency Management Agency administrator Craig Fugate was summoned to Congress to testify on the possible impacts of a cyberattack on the US electric grid. Water and wastewater systems are some of the first things to go down, he noted. “There is not really a good way to manage that if those systems go offline for extensive periods of time,” Fugate said.

He explained that the emergency response will become a game of triage: distributing enough power, gas, and generators to emergency services and utilities, while also trying to keep consumer-facing supply chains operating.

“Can you get enough life support and infrastructure going to keep the major supply lines up?” Fugate continued. “You are not going to have everything. You are not going to have what the normal consumption rates are.”

Lloyd’s estimates that the total economic costs and losses could hit $1 trillion.

The US Grid Attack Looming on the Horizon

Shift in cyberattack focus puts APAC region under growing pressure.

India, China and the US were the top DDoS attacks targets in Q1 2025, with APAC facing over half of global attacks, according to new data from cybersecurity firm StormWall.

A new report from cybersecurity firm StormWall reveals that nearly half of all DDoS attacks in the first quarter of 2025 were aimed at China, India and the United States. Instead of focusing heavily on the usual targets, attackers are now directing more of their efforts toward infrastructure in the Asia-Pacific region.

According to StormWall’s analytics center, India topped the chart with 18.1% of the global DDoS traffic, followed by China at 16.2% and the US at 14.7%. Japan (12.3%) and Taiwan (10.2%) rounded out the top five. When it comes to industry-wise attacks, Telecommunications became the top target in APAC, with 136% year-over-year (YOY) growth.

Four of the five most targeted countries now sit in the APAC region, with 57% of all malicious requests focused there. This shows that attackers are moving their focus away from the usual targets in Europe and North America, at least for now. The reasons may be geopolitical, economic, or a combination of both.

“You can always count on US targets attracting a lot of attention,” said Ramil Khantimirov, founder of StormWall. It’s the world’s largest GDP and hackers are always going to target that, but we’ve never seen so much activity in APAC and so densely concentrated.”

According to StormWall’s report shared with Hackread.com ahead of publishing on Thursday, the main force behind these attacks is botnets. These networks of infected devices were responsible for nearly 70% of all DDoS traffic tracked by StormWall in the first quarter of 2025.

Botnets, often run by state-sponsored groups, hacktivists, or organized cyber criminals, allow attackers to direct massive traffic at targeted systems. One such botnet identified in the report, named “Eleven11bot,” used a network of 86,000 compromised IP cameras and DVRs. It was frequently used in politically motivated campaigns.

While APAC faced an increase in DDoS attacks, other regions haven’t been spared. Belgium, though much smaller, experienced significant attacks, with 9.7% of global DDoS activity aimed at its national digital infrastructure, including MyGov.be and the Walloon Parliament. Saudi Arabia was also hit hard, at 6.8%, largely due to pro-Palestinian attacks linked to ongoing regional tensions.

Countries like Italy, Russia, and Switzerland appeared further down the list with lower attack volumes, but experts caution that even minor increases in activity can pose serious threats to unprepared networks.

Via StormWall

StormWall’s data paints a picture of a fast-evolving threat environment, driven by automation, political motives, and shifting strategic interests. While botnet-powered campaigns grow in scale, the need for mitigation services is more important than ever.

However, separate research from NordPass shows that many industries still fall short on basic cybersecurity. In most cases, weak and easily guessed passwords like 123456 and passwords@, are still in use, making systems vulnerable to breaches and turning devices into part of larger botnets.

StormWall Reveals India, China and US Faced Most DDoS Attacks in Q1 2025

kro (Kube Resource Orchestrator) 0.1.0 before 0.2.1 allows users (with permission to create or modify ResourceGraphDefinition resources) to supply arbitrary container images. This can lead to a confused-deputy scenario where kro's controllers deploy and run attacker-controlled images, resulting in unauthenticated remote code execution on cluster nodes.

Attack vector: More severe the more the remote (logically and physically) an attacker can be in order to exploit the vulnerability.

Attack complexity: More severe for the least complex attacks.

Privileges required: More severe if no privileges are required.

User interaction: More severe when no user interaction is required.

Scope: More severe when a scope change occurs, e.g. one vulnerable component impacts resources in components beyond its security scope.

Confidentiality: More severe when loss of data confidentiality is highest, measuring the level of data access available to an unauthorized user.

Integrity: More severe when loss of data integrity is the highest, measuring the consequence of data modification possible by an unauthorized user.

Availability: More severe when the loss of impacted component availability is highest.

GHSA-7633-x85h-5mqh: kro Confused Deputy vulnerability

Hewlett Packard Enterprise (HPE) has released security updates to address as many as eight vulnerabilities in its StoreOnce data backup and deduplication solution that could result in an authentication bypass and remote code execution.
"These vulnerabilities could be remotely exploited to allow remote code execution, disclosure of information, server-side request forgery, authentication bypass,

Latest News