Connect with us
DAPA Banner
DAPA Coin
DAPA
COIN PAYMENT ASSET
PRIVACY · BLOCKDAG · HOMOMORPHIC ENCRYPTION · RUST
ElGamal Encrypted MINE DAPA
🚫 GENESIS SOLD OUT
DAPAPAY COMING

Tech

AI agents show they can create exploits, not just find vulns

Published

on

AI + ML

Mythos and GPT-5.5 muscle out the competition

Sure, AI agents such as Mythos can find security vulnerabilities in software, but the bigger question is whether they can turn those flaws into functional exploits that work in the real world. After all, many AI-discovered bugs prove minor or difficult to weaponize. New research, however, suggests frontier models can indeed develop working exploits when directed to do so.

To better understand the rapidly changing security landscape, computer scientists from UC Berkeley, Max Planck Institute for Security and Privacy, UC Santa Barbara, Arizona State University, Anthropic, OpenAI, and Google decided to build ExploitGym, a benchmark for evaluating the exploitation capabilities of AI agents.

Advertisement

This is not an entirely disinterested set of investigators – Anthropic, OpenAI, and Google all sell AI services. And both Anthropic and OpenAI have talked up the risk of leading models Claude Mythos Preview and GPT-5.5 while selling access to government partners.

Since Anthropic announced Mythos in early April, the security community has been critical of the company’s approach, described by some as fear-mongering. And various security experts have made the case that even commercially available AI models can find security flaws.

Nonetheless, Mythos and GPT-5.5 outshine their peers in ExploitGym, as described in the paper, “ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?

ExploitGym consists of 898 real vulnerabilities found in applications, Google’s V8 JavaScript engine, and the Linux kernel. Its workout consists of presenting an AI agent with a vulnerability and proof-of-concept input that triggers it, to see whether the agent can create an exploit capable of arbitrary code execution.

Advertisement

According to the UC Berkeley Center for Responsible Decentralized Intelligence, Mythos Preview successfully exploited 157 test instances and GPT-5.5 managed 120 in the allotted two-hour window.

“Even when standard security defenses like ASLR or the V8 sandbox were turned on, a meaningful number of exploits still worked,” the boffins wrote in a blog post. “More strikingly, agents sometimes discovered and exploited entirely different vulnerabilities than the ones they were pointed at.”

The agents (CLI + model) tested were Claude Code with Claude Opus 4.6, Claude Opus 4.7, Claude Mythos Preview, and GLM-5.1; Codex CLI with GPT-5.4/GPT-5.5; and Gemini CLI with Gemini 3.1 Pro. And even the ancient models released in February (Opus 4.6 and Gemini 3.1 Pro) had some success.

Model benchmark comparison table showing agent success rates by category (userspace, browser V8, kernel), costs, and time across different AI models.

Model Agent Total U B K Cost (USD) Time (min)
Succ. Full Succ. Full
Claude Mythos Preview Claude Code 157 107 38 12 54.7 102.1
Claude Opus 4.6 Claude Code 15 12 2 1 8.08 21.76 18.1 66.7
Claude Opus 4.7 Claude Code 7 4 3 0 8.64 3.40 22.1 14.4
Gemini 3.1 Pro Gemini CLI 12 10 2 0 8.56 9.02 51.1 75.6
GLM-5.1 Claude Code 4 4 0 0 3.75 6.39 63.3 118.0
GPT-5.4 Codex CLI 54 38 15 1 12.20 25.43 51.1 103.5
GPT-5.5 Codex CLI 120 71 27 22 22.99 34.55 49.6 69.8

U = Userspace  ·  B = Browser V8  ·  K = Kernel  · 
Succ. = successful runs  ·  Full = full benchmark  · 
† preview model  ·  ‡ see notes

Advertisement

The researchers say that one of their more interesting findings is that these models sometimes went “off-script” in capture-the-flag (CTF) environments, where an agent has to find and retrieve some hidden value.

This was most evident with Mythos Preview and GPT-5.5. The former succeeded in 226 CTF exercises but only used the intended bug in 157 instances, while the latter captured 210 flags and only used the intended bug in 120 of those cases.

The authors also note that while there was some overlap in the exploits discovered, the various models found different exploits. This suggests applying a diverse set of models might be advantageous both in attack and defense scenarios.

It’s worth adding that ExploitGym tests were done with security guardrails disabled. When the test was re-run on GPT-5.5 with default safety filters active, the model refused 88.2 percent of the time before making any tool call. 

Advertisement

The Register, however, has seen security researchers craft prompts in a way to avoid triggering refusals. So safeguards of that sort have limits.

“Our results show that autonomous exploit development by frontier AI agents is no longer a hypothetical capability,” the authors state in their paper. “While current agents are not yet reliable across all targets, they already exploit a non-trivial fraction of real-world vulnerabilities, including complex targets such as kernel components.” ®

Source link

Advertisement
Continue Reading
Click to comment

You must be logged in to post a comment Login

Leave a Reply

Tech

Mozilla is fighting a losing battle to prove VPNs are essential privacy tools for everyone

Published

on

There was a time when VPNs were mostly associated with office workers logging into company servers or people trying to watch region-locked Netflix libraries. Now, they’ve somehow become one of the internet’s most politically sensitive tools.

Mozilla is the latest company sounding the alarm, warning UK regulators that VPNs remain “essential privacy and security tools” that should not be weakened or treated like suspicious circumvention software. The statement comes amid growing debates around online age verification systems, content controls, and broader internet regulation across Europe and beyond.

Mozilla says weakening VPNs could seriously hurt online privacy

In its latest policy push, Mozilla argues that VPNs are critical for protecting users from surveillance, cyberattacks, and invasive data collection. The company also warned regulators against creating laws or technical frameworks that indirectly discourage VPN usage or make them harder to access.

The timing here is important. Several governments are increasingly framing VPNs less as privacy tools and more as ways to bypass online restrictions, especially as countries roll out mandatory age-verification systems and tighter internet controls. Even the European Union has recently hinted at tougher scrutiny around VPN usage tied to online safety initiatives.

Mozilla’s concerns also come shortly after the company announced plans to expand privacy protections inside Firefox itself, including built-in VPN functionality as part of a broader browser overhaul.

The bigger problem is that VPNs are slowly becoming “suspicious” by default

The awkward reality is that Mozilla might already be fighting a losing battle here. Around the world, governments are becoming increasingly aggressive about restricting encrypted internet access and anonymous browsing tools. Countries like China, Iran, Russia, Iraq, and Myanmar already heavily restrict or outright ban VPN usage in various forms, while others are actively discussing tighter controls.

Advertisement

Ironically, the more governments try to clamp down on VPNs, the more essential they become for journalists, activists, remote workers, businesses, and even regular users simply trying to protect their browsing activity on public networks. In fact, in a recent conversation with Russian business outlet RBC, Valery Fadeyev, head of Russia’s Presidential Council for Civil Society and Human Rights, also admitted that fully banning VPNs is “simply impossible”.

“It became clear fairly quickly that this is an extremely complex system and that banning or switching off VPNs altogether is impossible.” – Fadeyev

That’s what makes this entire debate feel so messy. VPNs are simultaneously being treated as cybersecurity necessities and internet loopholes, depending on who’s talking. And honestly, once privacy tools start being framed as inherently suspicious, it becomes very hard to convince regulators otherwise.

Source link

Advertisement
Continue Reading

Tech

Extract 3D Video Game Content By Firing Up Photo Mode

Published

on

Here’s a pretty clever method [Dung3onlord] used to capture 3D scenes from a PlayStation 5 without needing any specialized software. All that’s needed is a series of high-resolution screenshots, and a few software tools.

The process is essentially photogrammetry, it just uses screenshots as the input instead of photographs.

Instead of sneakily yanking 3D assets from the runtime, he fires up the game’s photo mode on his PS5. By capturing an orbiting video of a static scene (making sure to hide the game’s user interface, something photo mode in games is good for) he ends up with a video file whose content — essentially a series of screenshots — can be used to reconstruct the original 3D scene. The workflow [Dung3onlord] uses has rather more steps, but conceptually that’s all there is to it.

The whole process is remarkably similar to photogrammetry, a method of turning a bunch of photographs from different angles into a 3D point cloud. We’ve seen photogrammetry used to digitize objects because point clouds can be turned into 3D models, essentially allowing one to 3D scan an object using little more than a digital camera.

In [Dung3onlord]’s case, once the point cloud is cleaned up and background removed, the scene is used to generate a gaussian splat which is then viewed through a VR headset.

Gaussian splats are especially well-suited to displaying colorful, organic 3D scenes that look pretty fantastic from any angle and are computationally simple to view. Want to see for yourself? [Dung3onlord]’s resulting scene is available to be viewed online.

Advertisement

It’s pretty cool stuff, but using photo mode as a way to capture game content, then reconstructing that content with tools intended for use with photos is an inspired solution. Be sure to check out the video overview of the process below.

Capture and view PS5 characters on a Quest (link and quick guide in the comment)
byu/Dung3onlord inOculusQuest

Source link

Advertisement
Continue Reading

Tech

Qualcomm’s New QCC74x Appears To Target The ESP32 MCUs

Published

on

These days wireless microcontrollers featuring built-in WiFi and Bluetooth are all the rage, with Espressif’s range of ESP32 MCUs being the default option for commercial and hobbyist projects alike. This makes Qualcomm’s recently released QCC74x MCU rather interesting, as specification-wise it would seem to be placed firmly in ESP32 territory.

On the radio side you get 1×1 WiFi 6, Bluetooth 5.4, and IEEE 802.15.4 (e.g. Thread and Zigbee), coupled with a single-core 352 MHz RISC-V CPU with FPU and DSP features and 484 kB of SRAM. The SDK for this MCU is hosted on Codelinaro, featuring the typical FreeRTOS-based stack, though confusingly Bluetooth and Zigbee support are currently marked as ‘not supported’. This might still be in progress.

Where the competition with Espressif feels clear is in the pricing, with the highest-performance evaluation board (QCC748M EVK, pictured above) listed for $13 (before taxes/tariffs). This gets you 8 MB of PSRAM built-in with unspecified link speed, but likely the same QSPI as used for the NOR Flash. USB support is available on this higher-end tier, while absent on the QCC743. Development documentation is also available, and looks fairly complete based on first glance.

Advertisement

Overall the QCC74x looks to be an upgrade to the older and significantly less powerful QCC730 MCU. Depending on software support and final pricing it could make for an interesting competitor to some of Espressif’s modules like its ESP32-C series or ESP32-S2, though the upcoming ESP32-S31 would seem to have it matched or beat on all metrics.

Source link

Advertisement
Continue Reading

Tech

How the Long Island truck hijackers were caught

Published

on

An arrest is made in the Morgan McSweeney stolen iPhone case, a guilty plea in the Beyonce stolen music case, and an AirTag helps recover a stolen Lamborghini, all in this week’s Apple Crime Blotter.

The latest in an occasional AppleInsider series, looking at the world of Apple-related crime.

Apple product thieves were caught after one activated stolen Apple Watches

On May 7, the Justice Department announced indictments of three people for their parts in the “brazen daytime robbery” in January. That robbery entailed the hijacking of a truck near a Long Island Apple Store, and the theft of $1.2 million of MacBooks, iPhones, iPads, Apple Watches, and accessories.

The New York Times offered some detail about how the suspects were caught.

Advertisement

Leads included fingerprints on paperwork, one of the accused thieves renting a storage unit in his own name, and another of them activating two stolen Apple Watches.

There has been an arrest in the highest-profile recent iPhone theft in Britain.

In October of 2025, a government-issued iPhone was stolen from Morgan McSweeney, who at the time was chief of staff for British Prime Minister Kier Starmer.

The loss of the iPhone meant that McSweeney could not produce messages between himself and Lord Peter Mandelson, the former British ambassador to the U.S., who was in hot water over his ties to Jeffrey Epstein. Mandelson was removed and later indicted, and McSweeney stepped down in February.

Advertisement

In late April, The BBC reported, a man was arrested for selling McSweeney’s stolen iPhone. The man is not believed to be responsible for the original theft.

Also, the phone itself has not been recovered.

Man pleads guilty to stealing MacBooks, Beyonce’s unreleased music

In the summer of 2025, a rental car belonging to a member of pop superstar Beyonce’s entourage was broken into in Atlanta. This led to the theft of hard drives and computers, including a MacBook Air, which contained the singer’s unreleased music.

An arrest was made in September, and just before a trial was set to begin, the man accused in the thefts pleaded guilty on May 12.

Advertisement

According to ABC 7, the 40-year-old defendant agreed to serve two years in prison followed by three more on probation. Before the plea was announced, prosecutors showed videos of the defendant approaching the vehicle and later driving it to an apartment complex.

The stolen items, per that report, included “two MacBook laptops, Apple headphones, luxury clothing and accessories,” and hard drives that investigators say contained unreleased Beyonce material.

Teens arrested for thefts of iPhone cases

A lot of thieves around the world, including in the U.K., steal iPhones. Two teenagers in that country were arrested in that country for stealing iPhone cases.

Eastern Daily Press reports the two teens were arrested in Attleborough, where officers say they were carrying 42 iPhone cases, worth 2,500 pounds, or about $3,382.

Advertisement

The cases, the reports said, were stolen from Apple Stores in Norwich and Cambridge.

Federal indictment for man accused of hacking into victims’ accounts on iCloud, other services

Federal prosecutors have announced the indictment of a Maryland man who they say accessed the private information of nearly 200 victims.

A 41-year-old employee of a Maryland medical system reportedly used “various cyber intrusion techniques” to access victims’ accounts with “online services such as Google Photos, iCloud Photos, Gmail, and Microsoft 365, and social media accounts.” The scheme, as announced by the Justice Department, lasted for eight years.

The man has been charged with two counts of unauthorized access to a protected computer and one count of aggravated identity theft.

Advertisement

AirTag helps in recovery of stolen Lamborghini, after chase

Police used an AirTag to recover a stolen 2021 Lamborghini Urus, valued at $300,000, in New Jersey in late April.

According to CBS New York, surveillance video caught thieves stealing the car from a custom auto body shop called MoeModz. However, the car’s owner had thought to place an AirTag inside the Lamborghini, which police were able to track.

This led to a chase that reached the Garden State Parkway, which ended when the car’s occupants got out and fled on foot. One of them, a 21-year-old, was charged with receiving stolen property and resisting arrest.

The Lamborghini suffered $15,000 in damage.

Advertisement

Couple follows AirPods to scrapyard to find stolen car

Another case of an owner using Apple technology to find a stolen car occurred in Milwaukee in early May.

TMJ 4 reported that a couple from that city had left their car on a street following a tire blowout. When they came back with a spare, the car was gone.

After calls to various towyards proved fruitless, the owner checked Find My iPhone and realized his AirPods had been left in the car.

Unfortunately, once they followed the signal, the car was in a scrapyard, “already crushed.” The Milwaukee Police Department is investigating the theft.

Advertisement

AirPods tracked after vehicle burglaries

In another case of AirPods being tracked to find stolen merchandise, two men were arrested after a series of car burglaries in Quad Cities, Iowa.

According to KWQC, there were more than 15 car break-ins between early March and early May, and after one, in which a credit card and AirPods were taken, police tracked the AirPods to an apartment building. There, they found the AirPods and other stolen items, leading to the arrests.

Two charged with series of Facebook Marketplace thefts

Two Miami men, aged 20 and 22, have been arrested and charged with a series of Facebook Marketplace thefts in which they’re accused of stealing iPhones, iPads, and MacBooks.

WSVN reports police say the two men arranged meetings to buy the products on Marketplace, and at the meetups, they would steal the items without paying.

Advertisement

One of the two faces charges of robbery by sudden snatching, strong-arm robbery, burglary of an occupied conveyance, and grand theft.

TV host in the Philippines gets stolen iPad back

Teddy Corpuz, a well-known TV host in the Phillippines, had his iPad stolen, among other items, in a break-in May. However, per GMA News Online, Corpuz announced on Facebook that he has recovered the iPad.

Among other stolen items were “cash, jewelry, [and] a charger.”

Advertisement

Source link

Continue Reading

Tech

Corsair Vanguard Air 99 Wireless review: a premium keyboard with great keys and a high price

Published

on

Why you can trust TechRadar


We spend hours testing every product or service we review, so you can be sure you’re buying the best. Find out more about how we test.

Corsair Vanguard Air 99 Wireless: One-minute review

The Corsair Vanguard Air 99 Wireless is an optical-mechanical hybrid board that takes aim at the premium end of the market.

It’s a smart and understated affair, although the bright RGB lighting certainly adds some vibrancy, as does the white colorway. The floating keycaps are also a nice touch.

Advertisement

Source link

Continue Reading

Tech

Doctors Explain Why Your Smartwatch Is Giving You Anxiety, and How to Stop It

Published

on

Whenever I wear a smartwatch, I find that my anxiety increases — specifically, my health anxiety. Also known as hypochondria or illness anxiety disorder, this type of anxiety makes me worry that I am or may become ill even when I’m healthy.

What’s ironic is that part of my job involves testing health-monitoring wearables, including fitness trackers and smart rings. While I love exploring this technology and do think it can help you learn more about your body, I have to be careful about how I use it so my anxiety isn’t triggered. I know I’m not alone

“Healthy adults and individuals with pre-existing medical conditions are increasingly using these devices to manage their health,” says Dr. Lindsey Rosman, assistant professor of medicine in the Division of Cardiology and co-director of the Cardiovascular Device and Data Science Lab at the University of North Carolina School of Medicine. “Whether 24/7 access to health information from a wearable actually helps or potentially harms people is really unclear.”

Advertisement

When you add in the ability to search your symptoms online or ask an AI chatbot in your wearable’s app every health question under the sun, it becomes even more difficult to discern between what’s helpful and harmful. 

To help myself and others with health anxiety navigate the world of wearables so we can either enjoy using them or know when it’s time to stop, I reached out to experts for their advice.

1. Turn off health alerts

Rosman has observed clinically that it can be beneficial to either scale back or turn off the features that make you anxious. This can be especially helpful for people with pre-existing conditions that are already being treated, such as atrial fibrillation (AFib, an irregular heartbeat), as your wearable’s irregular heart rhythm notifications will only make you anxious and can prompt you to see your doctor when it’s not medically necessary.

Plus, certain medications can affect the accuracy of wearable sensors, provoking false alarms. 

Advertisement

“We published a case report on a patient who performed over 900 EKGs [electrocardiograms or ECGs, which measure the heart’s electrical activity] on her smartwatch in a single year,” says Rosman. While most of the EKGs were normal, inconclusive alerts fueled her anxiety, leading to multiple ER visits, spousal conflict and the need for therapy to reclaim her daily life. The patient had no psychiatric history prior to getting a smartwatch.

An Apple Watch 11 showing the "Possible Hypertension" alert

When you get an unexpected health alert on your device, it can understandably cause panic.

Cole Kan/CNET/Apple

Dr. Karen Cassiday, author of Freedom from Health Anxiety and owner and managing director of the Anxiety Treatment Center of Greater Chicago, says that even patients who don’t have health anxiety can find wearables to be intrusive when they get too many alerts. “They discover they want to be less aware of every moment of their body’s functioning,” she says.

Advertisement

Thankfully, most wearable health features can be turned off completely or customized. 

For instance, Shyamal Patel, SVP of science at Oura, maker of the Oura Ring, shares that the device’s Personalized Activity Goals allow you to choose to see steps instead of calories, adjust your daily activity goal or hide calories completely, which can be necessary for anyone who finds calorie counting triggering or overly rigid. 

Advertisement

2. Avoid compulsively checking your device

Referring to a 2024 study she worked on that examined the impact of wearables on the psychological well-being of patients with AFib, Rosman says that about half of the participants were checking their heart rate every day out of habit, not because they felt symptoms. 

Cassidy explains that while people with health anxiety may initially find wearables helpful, compulsively checking to make sure their vitals are normal can accidentally become a form of negative reinforcement that further propels the anxiety.

“Often when I work with anxious people, we try to cut back or eliminate the need to compulsively check for reassurance on their wearables, as well as with ChapGPT or other digital ‘doctors,’” says Cassiday. 

When people refrain from compulsively checking, wearables can provide useful feedback that counters the false belief that something terrible will happen to their health.  

Advertisement

If checking your health metrics causes anxiety, try reducing how often you view them on your device or in its app. Setting an alert to check weekly, at a minimum, could help — especially since it’ll give you a broader picture, making you less likely to hyperfocus on a single data point that seems off. 

You should also avoid checking your wearable’s health information right after you wake up or before you go to bed, as this can set the tone for an anxious day or make it harder to fall asleep. 

If having a screen on your wrist makes it difficult for you to stop checking, a screenless smart ring or fitness tracker such as the Whoop 5.0 may be a better option, since they rely on apps instead of screens.

Advertisement
A close-up of the silver Oura Ring 4 on a pointer finger in front of a white wall.

A screenless smart ring may help you stop compusively checking your device.

Anna Gragert/CNET

“You choose how much or how little you engage with the app, which gives those who might be anxious about their health the option to limit the amount of time they spend with their data,” says Patel.

3. Focus on trends, not one-off metrics

When I asked both Patel and Dr. Jacqueline Shreibati, head of clinical for platforms and devices at Google, how people who wear their devices can reduce health anxiety, they emphasized the importance of tracking trends — not individual metrics.  

“We focus on long-term trends (rather than isolated metrics) to help users maintain a balanced relationship with their data,” says Shreibati. “What being healthy means differs for everyone, and we encourage users to consult their physician if they have any concerns.”

Advertisement

Patel points to the Tags and Trends features in the Oura app. Tags lets you tag lifestyle factors such as travel, alcohol, meditation or late meals, which you can then view in Trends to see how your behavior affects your recovery and sleep over weeks, rather than looking at a single score that may one day seem abnormal.

Sleet tracking Apple Watch Series 11

Instead of viewing a single sleep or stress score, consider looking at that data weekly or monthly.

Vanessa Hand Orellana/CNET

4. Remember: Your smartwatch can’t replace a doctor

“Most consumer wearables were originally developed as personal wellness devices, which are not required to demonstrate safety and efficacy like traditional medical devices (e.g., a blood pressure cuff or pacemaker),” Rosman explains. 

Advertisement

Yet we’ve begun using these wearables to monitor our health, using metrics such as heart rate and rhythm, blood oxygen, stress, sleep and physical activity. Now, some of these devices have medical-grade sensors, software and algorithms approved by the US Food and Drug Administration to detect irregular heart rhythms, hypertension and sleep apnea.

Despite FDA approval, wearables are simply not doctors, and they cannot provide medical diagnoses or treatment. That’s why it’s essential to understand what your device actually measures.

The ECG feature on many smartwatches is just one example of this. FDA-cleared as it may be, a single-lead ECG that only uses one electrode to record your heart’s electrical activity from your wrist is not the same as the 12-lead, hospital-grade ECG a cardiologist would use. 

While your wearable’s ECG can surface a potential symptom worth investigating with your doctor, it can’t replace a professional or their medical-grade equipment.

Advertisement
apple watch ultra 3 ecg

Performing an ECG on your smartwatch is not the same as having that same measurement taken in a doctor’s office.

Viva Tung/CNET/Apple

The gap is even wider for features including stress and sleep scores, which haven’t been clinically validated because there’s no one single gold standard to validate against. These numerical scores are calculated from bodily signals such as heart rate, temperature, movement and heart rate variability, which tend to correlate with your stress and sleep states. But the translation from raw signal to “your stress score is 74” is more of an educated estimate.

“What you’re seeing is a rough indicator of how your nervous system is functioning, not a medical diagnosis,” Rosman emphasizes.

Advertisement

Patel adds that not all physiological stress is inherently negative. “Some forms of short-term physiological stress can be healthy and adaptive,” he says. “That’s why we aim to pair data with in-app context and insights, so members can better understand what they’re seeing rather than receiving that information in a vacuum.” 

Nonetheless, when you don’t know exactly what your wearable is measuring, a “bad” stress or sleep score can seem scary when it isn’t necessarily a cause for alarm, but rather a sign that you may want to have a deeper conversation with your doctor.

5. Get your doctor’s thoughts

Just like you should talk to your doctor before starting a new medication or diet, you should get their thoughts on whether you could benefit from using a wearable.

“Education is probably the most underused tool we have,” Rosman says. 

Advertisement

When you don’t know what a healthy heart rate or ECG looks like, one seemingly atypical reading can send you into a panic. That’s why it’s essential to speak with your doctor so you understand your own baseline and if a wearable makes sense for your current health condition.

As a guide, Rosman provides the following questions you can ask your doctor:

Advertisement
  • What type of wearable should I use? 
  • How often should I check this data? 
  • What are healthy numbers for me? 
  • What do I do when I get an alert? 
  • When should I call the clinic or seek emergency care versus waiting? 

“A fast heart rate after climbing stairs is not the same as a dangerous arrhythmia, but without that context, a notification can feel terrifying,” Rosman adds. “So much wearable-related anxiety comes not from the data itself, but from not knowing what to do with it.”

6. Know when it’s time to remove that device and get help

When asked when someone should consider parting with their wearable or seeing a professional for health anxiety, Cassiday says that it’s similar to what many notice when they keep checking their smartphone for the next text, TikTok or other digital data.  

“If you find yourself interrupting pleasurable activities or your free time to check, or if you feel anxious about not checking, you have a problem,” Cassiday states. 

For instance, if you only stop thinking that you’ll have a heart attack when you check your wearable and see your resting heart rate. Or, put simply, if you only feel at peace after someone or something, such as a wearable reassures you that you’re in good health, it’s time to get professional support. 

Advertisement
An aerial view of a version with blonde hair, a yellow shirt and light-wash jeans talking to a therapist while on a gray couch.

If health anxiety is making it difficult for you to enjoy life, then it’s time to talk to a professional.

Constantinis/Getty Images

To find help, Cassiday recommends using the resources provided by the Anxiety and Depression Association of America or the International OCD Foundation, as health anxiety can be related to obsessive-compulsive disorder. 

7. Consider cognitive behavioral therapy 

When you have health anxiety, the gold standard for care is cognitive behavioral therapy. It involves exposure to health-related worries without any form of reassurance and learning to accept the uncertainty that comes with not knowing our future health status, manner of death or time of death.  

“People need to learn that all the vague symptoms that trigger their health anxiety are just normal variations of normal body functioning and aging,” Cassiday explains. “They have to reframe the symptoms they notice as nothing to examine, discuss or manage and instead trust the facts of their other evidence of good health.”

CBT can help you live in the present instead of spiraling into the anxiety-inducing “What if?” of the future.

Who should and shouldn’t use wearables

Wearables can be great for people who like tracking their fitness to motivate them toward their goals, or for patients and their care teams when medically necessary. Though they usually cost hundreds of dollars, wearables can be less expensive than medical tests. Some are even HSA- or FSA-eligible

Advertisement

“In AFib specifically, being able to correlate your symptoms with actual rhythm data can be genuinely empowering,” Rosman says. She’s observed that the patients who thrive with wearables are those who use the data as information — not as something to fear — and those who don’t participate in 24/7 surveillance.

In Rosman’s 2024 study, two-thirds of AFib patients said their wearable made them feel safer and more in control. Even so, there is still the risk of unintended consequences.

Two fitness tracker watches and a gold Oura Ring on a wrist and finger.

While they can be beneficial, wearables can also come with risks — especially since there isn’t enough research on the subject.

Advertisement

Giselle Castro-Sloboda/CNET

Just as doctors would never prescribe a medication without knowing the potential benefits, risks and how to manage them, wearables should be no different. “The technology has moved so much faster than the science, and we need the scientific evidence from clinical trials to catch up,” Rosman explains. 

Since the evidence isn’t there yet, Rosman is hesitant to say anyone should categorically avoid wearables. 

Despite that, people who are highly anxious about their heart or prone to obsessive symptom monitoring should approach with caution. The same goes for those with conditions involving unpredictable, abrupt symptoms, such as paroxysmal AFib and POTS, because the uncertainty of not knowing when the next episode will hit is stressful enough, and constant monitoring can make it worse.

A note on the science (or lack thereof)

Rosman has conducted research on the connection between wearables and anxiety, including a 2025 review describing the psychological effects of wearables on patients with cardiovascular disease and a 2024 study examining their impact on the psychological well-being of patients with AFib. 

The 2025 review found that while wearables can help promote healthy behaviors and provide data for diagnosis and treatment, they also pose risks, such as adverse psychological reactions. 

In the 2024 study, it was concluded that wearables were connected with higher rates of patients becoming preoccupied with their symptoms, being concerned about their treatments and using both formal and informal health care resources.

Advertisement

On the other hand, a 2021 study that analyzed the 2019 and 2020 US-based Health Information National Trends Survey found that using wearable devices for self-tracking can indirectly reduce psychological distress. Still, misinterpretation of wearable data may cause unnecessary panic and anxiety. 

A 2020 qualitative interview study featuring patients with chronic heart disease also found that while wearables’ data may be a resource for self-care, it can create uncertainty, fear and anxiety.

Ultimately, more studies are needed. 

“Honestly, we don’t have good scientific evidence in this area yet,” says Rosman. “Despite widespread use, there have been no clinical trials I’m aware of that have looked at the benefits and potential health risks of specific wearable health features.”

Advertisement

Rosman’s team plans to be the first to investigate this in patients with pre-existing heart conditions.

Wearables’ impact on our health care system

When wearables cause health anxiety, they can prompt healthy individuals to schedule unnecessary doctor’s appointments. This places a burden on our health care system, which is already experiencing shortages, making it difficult for people who actually require medical attention to access care. 

Rosman’s 2024 study found that those using a wearable sent nearly twice as many patient portal messages to their doctors. Responding to these messages from patients takes time, isn’t reimbursed by insurance and can contribute to burnout.

Advertisement
A person in blue scrubs with long brown hair checking messages on a desktop computer.

When health anxiety caused by wearables prompts people to message their doctors, it can put a strain on the health care system.

MoMo Productions/Getty Images

As a result, Rosman believes we need better systems for managing wearable data in clinical settings before we scale it further: “Wearables are changing how we deliver care in ways we haven’t fully prepared for.”

Wearables can further widen health care inequity due to their cost. 

“These devices are expensive, they were mostly designed and tested in young healthy people and they’re marketed toward higher-income consumers,” Rosman explains. “If we’re not thoughtful about access, wearables could actually widen health disparities rather than close them. That’s the opposite of what we want.”

The bottom line

While wearables have their benefits, there are also risks to consider, especially given the limited research on the subject.

If you purchase a wearable and it triggers health anxiety, you don’t have to use every available feature, wear it constantly or continue to wear it at all. Before you even buy that device, you can arm yourself with anxiety-reducing knowledge by getting your doctor’s expert opinion.  

Advertisement

However, if health anxiety continues to take over your life, it may be time to remove your wearable and seek professional help. 

As for me, writing this piece has been a necessary reminder that, while there’s a lot we can’t control in life, the power is in our hands (or on our wrists or fingers) when it comes to the technology we put on our bodies or invite into our homes. Just like an itchy sweater or a lumpy armchair, we can send the technology that doesn’t serve us packing.  

Source link

Advertisement
Continue Reading

Tech

Gemini Intelligence has strict requirements, and your phone may not qualify

Published

on

Google’s new Gemini Intelligence platform is quickly becoming one of the biggest talking points in the Android world right now. After being highlighted during this week’s Android Show, the feature is already being tied to several upcoming premium foldables and flagship phones. But there’s a catch: not every high-end Android device will be able to run it. And surprisingly, even some of Google and Samsung’s latest foldables may miss out.

According to Google’s requirements, Gemini Intelligence isn’t just another software update you can casually push to older devices. The company appears to be building this around a much stricter hardware and long-term software support system. To qualify, a phone needs a flagship-grade chipset, at least 12GB RAM, support for AI Core, and Gemini Nano v3 or newer. That immediately creates a problem for several current-generation phones.

Gemini Intelligence needs more than just a powerful chip

Google’s requirements go beyond raw performance. Devices also need to promise at least 5 Android OS upgrades and 6 years of security patches, with quality standards tied to system stability and crash rates.

While many flagship phones already offer long software support cycles, the Gemini Nano version requirement seems to be the real barrier here. Reports suggest devices like the Pixel 9 series and Samsung’s Galaxy Z Fold 7 are still running Gemini Nano v2, meaning they don’t currently qualify for Gemini Intelligence support.

Advertisement

The feature list is expected to expand significantly across 2026 Android flagships, including the Pixel 10 series and the Oppo Find X9 lineup, which are likely being designed with these AI requirements in mind from the start.

That said, the situation is still slightly unclear. Google’s documentation specifically mentions support for Gemini Nano’s Prompt API rather than directly confirming whether older devices are permanently excluded. So there’s still a possibility that some phones could gain compatibility later through future Android updates or backend upgrades.

The RAM requirement could reveal Google’s bigger AI plans

One of the more interesting details here is Google’s insistence on a minimum of 12GB of RAM for Gemini Intelligence. That’s a fairly aggressive requirement, especially given that some leaks have suggested the base Pixel 11 might actually ship with only 8GB of RAM. If these new AI requirements are accurate, those earlier leaks may not tell the full story.

It would be odd for Google to heavily market advanced on-device AI features while simultaneously lowering memory capacity on its own flagship phones. For now, Google says Gemini Intelligence will first arrive on Pixel and Samsung Galaxy devices later this year.

Source link

Advertisement
Continue Reading

Tech

Research repository ArXiv will ban authors for a year if they let AI do all the work

Published

on

ArXiv, a widely used open repository for preprint research, is doing more to crack down on the careless use of large language models in scientific papers.

Although papers are posted to the site before they are peer-reviewed, arXiv (pronounced “archive”) has become one of the main ways that research circulates in fields like computer science and math, and the site itself has become a source of data on trends in scientific research

ArXiv has already taken steps to combat a growing number of low-quality, AI-generated papers, for example by requiring first-time posters to get an endorsement from an established author. And after being hosted by Cornell for more than 20 years, the organization is becoming an independent nonprofit, which should allow it to raise more money to address issues like AI slop

In its latest move, Thomas Dietterich — the chair of arXiv’s computer science section — posted Thursday that “if a submission contains incontrovertible evidence that the authors did not check the results of LLM generation, this means we can’t trust anything in the paper.” 

Advertisement

That incontrovertible evidence could include things like “hallucinated references” and comments to or from the LLM, Dietterich said. If such evidence is found, a paper’s authors will face “a 1-year ban from arXiv followed by the requirement that subsequent arXiv submissions must first be accepted by a reputable peer-reviewed venue.”

Note that this isn’t an outright prohibition on using LLMs, but rather an insistence that, as Dietterich put it, authors take “full responsibility” for the content, “irrespective of how the contents are generated.” So if researchers copy-paste “inappropriate language, plagiarized content, biased content, errors, mistakes, incorrect references, or misleading content” directly from an LLM, then they’re still responsible for it. 

Dietterich told 404 Media that this will be a “one-strike” rule, but moderators must flag the issue and section chairs must confirm the evidence before imposing the penalty. Authors will also be able to appeal the decision.

Recent peer-reviewed research has found that fabricated citations are on the rise in biomedical research, likely due to LLMs — though to be fair, scientists aren’t the only ones getting caught using citations that were made up by AI.

Advertisement

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Source link

Continue Reading

Tech

How to watch Sinner vs Ruud: Free Italian Open 2026 Final Live Streams

Published

on

  • Jannik Sinner vs Casper Ruud: Sunday, May 17
  • Not before: 11am ET / 4pm BST / 1am AEST (Sun)
  • FREE stream: Rai Play (Italy)
  • Access your usual streaming services with NordVPN

Watch Jannik Sinner vs Casper Ruud live streams, as the Italian Open 2026 men’s final brings together arguably the two best clay courters on tour. Sinner, the overwhelming favorite, is going for another slice of history but Norwegian Ruud has his sights on the biggest of upsets.

Sinner is now one match from a career Golden Masters – only Novak Djokovic has won each of the nine Masters 1000 events in history. The 24-year-old Italian is also seeking to become the tournament’s first home men’s winner in 50 years and, having won his last 33 Masters matches, will start as a wide favorite.

Source link

Continue Reading

Tech

Classic 7 is Windows 10 LTSC cosplaying as Windows 7

Published

on

oSes

Uncanny rebuild resurrects the 2009 desktop, complete with support, updates, and licensing questions

For those who miss what Windows looked like in 2009, Classic 7 is a heavily modified version of Windows 10 IoT LTSC, reworked to make it look as much as possible like Windows 7, while still being in support and receiving updates.

This has been
accomplished thanks to a large compilation of skins, themes, add-ons,
tweaks, and so on – some of which are real components from older
versions of Windows, adapted and modified to run on Windows 10.

Advertisement

We were not sure whether to cover Classic 7, because while it is impressive and fun, we are not at all sure it is legitimate to use. But we can see a target audience.


The good old sky-blue login screen with its decorative vine tells you that things are not what they seem

This isn’t just a layer of makeup; it’s more like a face transplant.
It includes some real binaries from Windows 7, and indeed earlier
versions, adapted and grafted onto Windows 10. One component is the
Windows Media Center from Windows XP, which was cut
from Windows 10
before release.

The specific version of Windows 10 that it’s modified is significant.
It’s Windows 10 IoT LTSC. We talked
about this specific edition in April 2025
because it’s the last
version of Windows 10 that is still in support and receiving updates. The standard Windows 10 Enterprise LTSC release will continue to receive updates until 2027,
and the IoT edition, which is only available in US English, will get
updates until 2032 – so this is the longest-lived version of Windows 10.

At the bottom of our story on Windows 10 LTSC, we mentioned the
slightly shady world of third-party modified editions of Windows.
Classic 7 is one; it’s a modified version of an Enterprise edition of
Windows, one that’s only available for legitimate licensing via a Volume
License Agreement. Unless you have appropriate volume licensing for the underlying Windows edition and have paid the fairly hefty
fee, this is an unlicensed copy of Windows. So we have to spell out that
this is not for production use, and you should not use it in
any working environment. It’s an interesting hack, though, and it might
be a bit of fun for a home gaming machine or something like that. 

Advertisement

As an
aside, one of the most widely used tools for activating unauthorized
copies of Windows and Office, MassGrave, is in fact hosted
on GitHub
. In other words, Microsoft itself is hosting tools to
activate unlicensed copies of Windows and Office. Whether that counts as
tacit approval, we wouldn’t like to say.


This, unbelievably, is a Windows 10 desktop. Yes, we know it says it’s Windows 7 Ultimate

Classic 7 has been under construction for over a year and a half, and
it’s the sequel to an earlier project called Reunion7 – also hosted on GitHub, as it
happens.

As its list of
credits
shows, Classic 7 is in part a compilation of a lot of
existing tools. Some of them are relatively well known, such as Winaero Tweaker, which can run on any copy of Windows and, among lots of other options, allows some of the
less desirable changes in the Windows UI to be undone – for instance,
switching to the hidden Aero Lite theme.

Classic 7 includes this and a lot more besides. We could identify some of the couple of dozen credited projects, such as the Aero11 theme, itself a port
of Aero10 to Windows
11. This works alongside OpenGlass, which
brings Aero-style transparency to Windows 10.

Advertisement

There’s also the Windows NT
Modding Utility
, and another hack that lets you change the Windows
version number reported on the command-line, called Custom CMD
Version Text
. Multiple sub-components come from the Windhawk mods collection, some
credited to a developer called ImSwordQueen, whose themes
can be seen on
DeviantArt
.


Classic 7 runs the original Windows 7 Explorer, and there’s a README file on the desktop with credits

Other components are more than just cosmetic. For instance, the
remarkable description of Explorer7:
“explorer7 is a wrapper library that allows Windows 7’s
explorer.exe to run properly on modern Windows versions,
aiming to resurrect the original Windows 7 shell experience.” So this is not merely a theme for Windows 10 Explorer: as far as we can tell,
it’s the real Windows 7 Explorer, but running on top of 10. The same appears to apply to Control Panel as well, thanks to the Control
Panel Restoration Pack
. Thanks to the Windows
Media Center (Modern Hardware)
effort, this is the real XP version,
which an on-screen message says replaced the Windows 8 version used in
an older build.

We tried Classic 7 in VMware, and the experience is quite uncanny. We
did hit some glitches: our first installation failed when we let it do
its own disk partitioning. Deleting all the partitions, manually creating a single large C: drive, and telling the installer to use that worked. A few error messages did
appear here and there. Trying to change screen resolution went badly
awry until we installed the VMware guest additions. Opening Windows
Update just threw an error.

Overall, though, it is genuinely remarkable. It looks and feels
like Windows 7 – but in principle, you can run the latest apps and drivers and
they should work. It even includes your choice of older Firefox versions,
including version 115 ESR, skinned to look exactly like Internet Explorer – an effort called BeautyFox.

Advertisement

Although the menu says ‘About Internet Explorer,’ it isn’t: this is Firefox, and the title bar shows that alpha-blending is working

Last year, we wrote a piece on running
Windows 7 in 2025
and it really reminded us how great the
2009 release looked compared to anything that’s come since. Apparently,
that late-noughties translucent look is now known as Frutiger
Aero
, and frankly we miss it.

In all honesty, we feel Classic 7 goes too far. We don’t want
Help/About dialog boxes, and even the winver tool and
the ver command to lie to us. We’d prefer something that
told the truth, but looked pretty while doing it.

But as we wrote last year, some personal friends are still running
Windows 7 by choice, and compatibility is starting to become a
problem. If you want a recent Firefox, well, you’re out of luck. Firefox
115
from 2023 still works, and remarkably, it’s still
getting security fixes
now: the March end-of-life has been postponed
again, and it’s currently
August 2026
. The Irish Sea wing of Vulture Towers is still running
it on OS X 10.13 and it works flawlessly.

This is a way out: to keep the 17-year-old vintage look, while
running a codebase that still has another five years in it. If you’re
that determined, it’s an option… and it’s undeniably an attractive GUI.
Whether this unauthorized rebuild of an unlicensed OS is an attractive
option, though – you must decide that for yourself. ®

Advertisement

Source link

Continue Reading

Trending

Copyright © 2025