The idea that cancer could behave like a contagious disease may seem absurd; in humans, cancer cannot be transmitted from one person to another. Yet nature is still full of surprises. In exceptional cases, cancer cells themselves can pass from one individual to another and continue to multiply in a new host. This rare phenomenon has been detected in animals such as dogs, Tasmanian devils, and some species of mollusks, and it seems we can add one more to that exclusive club: catfish.

An international team of researchers has identified a transmissible melanoma in the brown catfish (Ameiurus nebulosus)—the first documented case in fish and also the first recorded in a freshwater ecosystem. Their findings were reported this month in the journal Nature.

An Impossible Pattern

The story began in 2012, when fishermen and biologists detected an unusually high number of catfish with black lesions in Lake Memphremagog, a body of water straddling Vermont and Quebec. Previous studies revealed that these spots were malignant melanomas, and that between 23 percent and 37 percent of the specimens examined had these tumors—an unusually high incidence for this species.

“That was surprising,” said Julie Dragon, a researcher at the University of Vermont and co-lead author of the study. “We wanted to know how a bottom-dwelling fish was getting a cancer we associate with exposure to too much sunlight.”

Initially, it was thought that the cause might be some toxic substance or a pathogen linked to pollution. One suspicion was that the flooding caused by Tropical Storm Irene in 2011 had degraded water quality by washing environmental pollutants into the lake. There were also concerns about the health of the lake, which supplies drinking water to more than 175,000 people. However, initial genetic analyses began to point in an unexpected direction.

It was then that the researchers wondered whether the tumors were descended from a single original cancer that had learned to spread among individual fish, rather than arising independently in each fish.

To test this unusual hypothesis, the team sequenced the complete genomes of tumors and healthy tissues from affected fish, and compared the data with that from healthy specimens from different populations.

The results indicated that the tumor cells from different fish were much more closely related to one another than to the animals in which they developed. In other words, the tumors shared their own genetic identity, distinct from that of their hosts. Furthermore, hundreds of thousands of genetic variants appeared repeatedly in the tumors but were absent from the healthy tissues of the same fish.

Such a pattern would be virtually impossible if each melanoma had arisen independently. By way of comparison, the researchers analyzed hundreds of human melanomas and found that nearly all of their mutations were unique to each patient. In contrast, most of the mutations present in the tumors of these fish were shared by numerous individuals—a hallmark of a transmissible clonal cancer.

The big picture: AMD had a busy week, announcing the Ryzen 7 7700X3D CPU for gaming PCs and the Epyc Venice lineup for agentic AI workloads in data centers and high-end workstations. The company also launched the Helios rack-scale AI system at its Advancing AI 2026 event in San Francisco and previewed its next-generation Zen 7 and Zen 8 CPU architectures.

In her keynote speech, AMD CEO Lisa Su revealed that Zen 7 will debut in 2028 with the 7th-generation Epyc “Florence” processors for enterprise and AI data centers. The new core architecture will be divided into classic (Zen 7) and compute-optimized (Zen 7c) versions, with both expected to be manufactured on TSMC’s cutting-edge A16 or A14 process node.

AMD did not reveal any architectural details about Zen 7, but confirmed that Epyc Florence will include both Zen 7 and Zen 7c cores. The Florence processors will also be the first processors in AMD’s portfolio to support the next iterations of MRDIMM and LPDDR memory standards, as well as ACE AI Compute Extensions to accelerate matrix multiplication and AI workloads across the x86 ecosystem.

Epyc Florence will cover the existing SP7 and SP8 platforms, with the SP7 chips delivering high-core-count and flagship performance, while the SP8 versions will offer optimized performance with a focus on value. Both versions will power AMD’s next-gen AI rack solution, “Ferrara,” which will be optimized for high-performance AI host nodes.

Moving beyond Zen 7, AMD added that it has started work on its Zen 8 core architecture, which will debut in 2030 with the company’s 8th-generation Epyc chips, codenamed “Ravenna.” The company did not reveal anything else about either Zen 8 or Epyc Ravenna, including architectural details, process node, core count, or memory support.

Team Red also unveiled the Instinct MI455X AI accelerator with CDNA 5 compute architecture and confirmed that the CDNA 6-based MI500 will debut in 2027, followed by the MI600 in 2028. The company has yet to confirm the CDNA generation, memory configurations, and compute specifications for the MI600.

Moving to the Helios AI rack roadmap, the first-generation models are set to launch this year, powered by Epyc Venice CPUs and the new MI455X accelerator. Helios 500 is slated to arrive in 2027 with Epyc Verano CPUs and Instinct MI500 GPUs, followed by Helios 600 in 2028, powered by Epyc Ferrara and Instinct MI600.

Earlier this week, AMD launched its latest 3D V-Cache CPU, the Ryzen 7 7700X3D. Aimed at gamers on a budget, the new processor comes with 8 Zen 4 cores, 16 threads, and 96MB of L3 cache. The company also unveiled the Zen 6-based Epyc Venice CPUs and revealed key details about its Arm-based Vera CPU, designed specifically for agentic AI workloads in data centers.

The big picture: AMD had a busy week, announcing the Ryzen 7 7700X3D CPU for gaming PCs and the Epyc Venice lineup for agentic AI workloads in data centers and high-end workstations. The company also launched the Helios rack-scale AI system at its Advancing AI 2026 event in San Francisco and previewed its next-generation Zen 7 and Zen 8 CPU architectures.

In her keynote speech, AMD CEO Lisa Su revealed that Zen 7 will debut in 2028 with the 7th-generation Epyc “Florence” processors for enterprise and AI data centers. The new core architecture will be divided into classic (Zen 7) and compute-optimized (Zen 7c) versions, with both expected to be manufactured on TSMC’s cutting-edge A16 or A14 process node.

AMD did not reveal any architectural details about Zen 7, but confirmed that Epyc Florence will include both Zen 7 and Zen 7c cores. The Florence processors will also be the first processors in AMD’s portfolio to support the next iterations of MRDIMM and LPDDR memory standards, as well as ACE AI Compute Extensions to accelerate matrix multiplication and AI workloads across the x86 ecosystem.

Epyc Florence will cover the existing SP7 and SP8 platforms, with the SP7 chips delivering high-core-count and flagship performance, while the SP8 versions will offer optimized performance with a focus on value. Both versions will power AMD’s next-gen AI rack solution, “Ferrara,” which will be optimized for high-performance AI host nodes.

Moving beyond Zen 7, AMD added that it has started work on its Zen 8 core architecture, which will debut in 2030 with the company’s 8th-generation Epyc chips, codenamed “Ravenna.” The company did not reveal anything else about either Zen 8 or Epyc Ravenna, including architectural details, process node, core count, or memory support.

Team Red also unveiled the Instinct MI455X AI accelerator with CDNA 5 compute architecture and confirmed that the CDNA 6-based MI500 will debut in 2027, followed by the MI600 in 2028. The company has yet to confirm the CDNA generation, memory configurations, and compute specifications for the MI600.

Moving to the Helios AI rack roadmap, the first-generation models are set to launch this year, powered by Epyc Venice CPUs and the new MI455X accelerator. Helios 500 is slated to arrive in 2027 with Epyc Verano CPUs and Instinct MI500 GPUs, followed by Helios 600 in 2028, powered by Epyc Ferrara and Instinct MI600.

Earlier this week, AMD launched its latest 3D V-Cache CPU, the Ryzen 7 7700X3D. Aimed at gamers on a budget, the new processor comes with 8 Zen 4 cores, 16 threads, and 96MB of L3 cache. The company also unveiled the Zen 6-based Epyc Venice CPUs and revealed key details about its Arm-based Vera CPU, designed specifically for agentic AI workloads in data centers.

Building your system with some extra capacity makes managing it a bit easier.

When does the 20 percent buffer come in handy?

Eoneren/Getty Images

Why should I spend more on solar if I don’t always need it?

Jeremy Poland/Getty Images

Does the extra 20 percent really make a difference?

Kangestudio/Getty Images

Nvidia will invest $1 billion in South Korean internet company Naver to help finance an AI data center under construction in South Korea, the chipmaker announced late Friday. The funding will allow Naver to more than triple the size of the facility from 55 megawatts to 200 megawatts, with US private equity firm Brookfield agreeing to a nonbinding term sheet to fund up to $9 billion for the project. The site will run on Nvidia’s AI computing hardware.

Separately, Nvidia announced a long-term commercial partnership with SK Group that it described as worth more than $500 billion. The figure includes money Nvidia will spend buying memory chips from SK Hynix, the world’s largest supplier of high-bandwidth memory, as well as purchases by SK Group of Nvidia’s supercomputers. SK Telecom will build more than 2 gigawatts of AI data centers on the Korean Peninsula, with the first facility due to open next year.

The announcements coincide with a visit to Silicon Valley by South Korean President Lee Jae Myung, and additional tech deals between Korean and American companies are reportedly in the works. Nvidia CEO Jensen Huang told Bloomberg Television that South Korea is in a “golden age,” pointing to the country’s booming semiconductor and industrial businesses as reasons to invest. Nvidia will also help SK Hynix design future high-bandwidth memory chips, an effort to guarantee access to the component that remains in short supply because of the global buildout of AI data centers.

The Korea deals are the latest in a year-long investment spree in which Nvidia has committed more than $40 billion to AI equity positions, including $30 billion in OpenAI. Earlier this week the chipmaker signed a separate agreement with Amkor Technology to bolster chip packaging, and it recently completed a trip to Japan that included deals with Toyota and Japanese robotics firms. Critics have described the pattern as circular financing, in which Nvidia invests in companies that then buy its hardware, a structure the chipmaker has replicated with data center operators, neoclouds, and now sovereign AI partners.

For South Korea, the deals represent a bet that its semiconductor industry can remain central to the AI buildout even as geopolitical tensions reshape chip supply chains. Samsung and SK Hynix are the only two companies in the world capable of producing the advanced memory chips that Nvidia’s processors require, and locking in a design partnership with Nvidia gives SK Hynix a structural advantage over its rival. Huang has argued that every country should create its own sovereign AI capability, and South Korea, with its chipmaking expertise and abundant engineering talent, is among the best positioned to do so.

Steam discussion forums are being abused in ClickFix attacks that pretend to be fixes for game and computer problems but actually infect devices with cryptominers.

BleepingComputer learned of the campaign from a reader, who told us threat actors are creating random Steam accounts to post what appears to be helpful fixes for people’s posts about games crashing, lost inventory items, and other technical issues.

The threat actors reply to posts, telling other members to open PowerShell as an administrator and run a command to fix the issue. However, when executing the command, it quietly downloads an XMRig miner executable and launches it on the computer.

These types of attacks are known as ClickFix, a social engineering tactic that displays fake errors, verification prompts, or troubleshooting instructions to trick victims into manually executing malicious commands.

Although ClickFix attacks require interaction from the victim, they are effective because they present users with what appears to be both a legitimate solution for a problem they are having.

Since the victim manually launches the command, the attack can also bypass some security protections that would otherwise automatically block executed malicious code.

Fake Windows optimization installs malware

The PowerShell script distributed in the Steam campaign masquerades as a Windows optimization utility named “msf utility \ PC Opt.”

When launched, it displays messages claiming to perform numerous maintenance tasks, including cleaning temporary files, flushing the DNS cache, updating drivers, checking the disk, turning off unnecessary startup items, scanning for malware, repairing the Windows image, and running System File Checker.

However, most of these functions do not perform the tasks they claim. Instead, they display fake progress messages and pause for a random period between 1.5 and 8 seconds to make the utility appear legitimate.

The actual malicious activity is hidden in a function named ‘Advanced-Optimization’, which first disables TLS certificate validation and checks that it is running with administrator privileges. If it is not, the script displays an error stating that administrator rights are required and exits.

Once running with elevated privileges, the script creates the ‘C:\Windows\Background’ directory and adds it as an exclusion to Microsoft Defender’s scanning functions.

It also attempts to stop an existing scheduled task named ‘XMRig-[computer name]’ and terminates matching processes named ‘xmrig’ or ‘system’ that are running from the installation directory. It also attempts to delete any XMRig configuration files stored as C:\Windows\Background\config.json.

It is unclear whether this cleanup is intended to remove leftovers from an earlier install of the same malware or another miner already present on the device.

The malware then creates a temporary outbound Windows Firewall rule allowing connections to ‘msfconfig[.]icu‘ over TCP port 443 and downloads the XMRig miner payload from https://msfconfig[.]icu:443/tmp/system.txt into a randomly named temporary file.

Before installing it, the script verifies that the downloaded file is not empty and is a valid executable. If so, it moves the file to C:\Windows\Background\system.exe.

So it launches every time Windows starts, it will create a new scheduled task named “XMRig-[computer name],” that launches the system.exe executable with SYSTEM privileges.

As a general rule, users should never run PowerShell commands provided by unknown users in discussion forums, even when the commands are presented as fixes for a problem they are currently experiencing.

Those who executed the command should check for the ‘C:\Windows\Background’ directory, a Microsoft Defender exclusion for that path, and a scheduled task beginning with ‘XMRig-‘.

If these signs of compromise are detected, you should immediately run an antivirus program to scan for malware and remove anything it finds.

If it does not detect the miner, you should manually stop and remove the XMRig-[computer name] scheduled task, remove the Microsoft Defender exclusion for C:\Windows\Background, and delete the folder and its contents.

Ultimately, it may be safer to reinstall the operating system, as there is no way to know whether the downloaded payload performed additional malicious actions while it was running.

Security teams log 54% of successful attacks and alert on just 14%. The rest move through your environment unseen.

The Picus whitepaper shows how breach and attack simulation tests your SIEM and EDR rules so threats stop slipping by detection.

Get the whitepaper

Modern electric vehicles retain an average of 97 percent of their original range after three years and 95 percent after five years, according to a new analysis by EV data company Recurrent. For a 2026 model with a rated range of 325 miles, that means roughly 309 miles would remain after half a decade of driving. A separate Recurrent study found that the battery replacement rate for EVs built from 2022 onward is just three in every thousand vehicles.

A study from vehicle data company Geotab, drawing on more than 22,700 vehicles, reached a similar conclusion, finding an average annual degradation rate of just over two percent. Viet Nguyen-Tien, an EV researcher at the London School of Economics, told The Wall Street Journal that improvements in battery chemistry, thermal regulation, and battery-management systems have made today’s batteries far more durable than those in early-generation models like the Nissan Leaf, which developed a reputation for rapid degradation after its 2010 launch. In a 2026 AAA survey, 56 percent of people who said they were unlikely to buy an EV cited the high cost of repairing or replacing the battery as their top concern.

The data suggests that concern is increasingly unfounded, but it has not translated into sales. Cox Automotive estimates that Americans bought roughly 75,000 new EVs in June, down nearly 28 percent from a year earlier, giving electric vehicles just over five percent of total new-vehicle sales during the month. A Deloitte survey from earlier this year found that only seven percent of Americans want their next car to be electric, and used EV sales told a different story, rising more than 20 percent in June compared with the previous year, according to Cox Automotive.

The sales decline is largely a policy story. The Trump administration allowed the $7,500 federal EV tax credit to expire last September and rolled back emissions regulations, a shift that Ford CEO Jim Farley warned at the time could cut demand for EVs in half. Several automakers have since cancelled or paused EV models in the US market, and while Edmunds reported that rising gas prices had started nudging consumer interest upward earlier this year, the shift was not enough to reverse the broader trend.

The gap between what the technology can do and what consumers believe it can do is now the central obstacle for the American EV market. The batteries work, the replacement rates are negligible, and the used market is growing, but new sales are falling because federal policy removed the price incentive that made EVs competitive with combustion alternatives. Until that changes, the industry is left selling a product that has answered its biggest technical question to a market that either has not heard the answer or no longer has the financial reason to care.

Motorola has revealed the Moto Pad 70, a new mid-range Android tablet with a large display, stylus support, 5G connectivity, and a sizeable battery.

The tablet will launch in India on August 8 through Flipkart. Its 12.1-inch 2.5K display, included Moto Pen, 10,200mAh battery, and 68W charging give it several features that could appeal to people considering Apple’s base iPad. However, the mid-range Dimensity 6400 processor means it is unlikely to match the iPad in raw performance.

Could it be a cheaper iPad alternative?

Motorola appears to be competing on value rather than processing power. The 12.1-inch screen is larger than the display on Apple’s standard iPad, while support for up to 800 nits of brightness and wide DCI-P3 colour coverage should make it suitable for streaming, reading, and creative work.

The included Moto Pen adds to the tablet’s value. It supports pressure sensitivity, tilt detection, and palm rejection, giving buyers a ready-made setup for note-taking and drawing without another accessory purchase. Motorola will also offer a magnetic keyboard case for writing and lighter office work.

A 10,200mAh battery should help it last through long viewing or study sessions, while 68W charging is unusually fast for a tablet in this class. Motorola has managed to fit it into a 6.29mm body weighing 530 grams, so the large battery has not made it especially thick or heavy. Quad speakers with Dolby Atmos and 5G connectivity add to its appeal as an entertainment device.

Performance is where the Pad 70 lags behind

The Dimensity 6400 should be sufficient for browsing, streaming, video calls, note-taking, and everyday multitasking. It will struggle to compete with Apple’s chips in demanding games, video editing, creative apps, and heavier workloads.

Motorola is also promising only two major Android upgrades, although security updates will continue until 2030. The Moto Pad 70 closely resembles the Lenovo Idea Tab Plus, which launched last September for around 300 euros. Indian pricing is often lower than a direct conversion would suggest, so Motorola could position it comfortably below the base iPad.

At the right price, its large screen, included stylus, 5G support, and fast charging could make the performance trade-off easier to accept. Currently, it is unclear if the company plans to launch this tablet outside of India.

A hot potato: After introducing Windows 11, Microsoft forced the entire PC ecosystem to turn the Trusted Platform Module (TPM) into a standard component for every new motherboard or CPU. Now, Redmond is exploiting the now-pervasive device to further strengthen the Windows activation process in the enterprise market.

Microsoft recently announced a brand-new addition to Key Management Service (KMS), a standard feature for mass activation of Windows devices in enterprises and other large-scale organizations. The KMS feature will soon rely on hardware-based security, using TPM’s encryption “brain” to verify the legitimacy of the server hosting KMS data.

Redmond explained that attackers have traditionally abused KMS to spoof the activation process, which is both a security issue and a way for individual users to avoid paying for a new Windows license. The new “TPM-based attestation” option will use the TPM for verifying the cryptographic proof of the integrity of a KMS server.

TPM-based attestation will first check the hardware identity of the KMS host, proving that the server has been verified by Microsoft as a legitimate hardware device. Then, the feature will confirm that the KMS host has not been tampered with in any way. Finally, the TPM chiplet will let the now-verified KMS host manage the mass activation requests required by the organization.

TPM-based attestation will become a mandatory requirement for KMS activation starting with the next Windows Server release, Microsoft stated. The corporation will begin to pressure business customers by sharing its “readiness messaging” in Windows Server 2025, starting from August 2026. Sysadmins will need to adapt and prepare their organizations in time, checking whether their KMS infrastructure is ready for hardware-based activation security.

“As Windows security continues to evolve, trusted activation infrastructure will play an increasingly important role,” Microsoft stated. “KMS Hardware Secured helps position your environment for the future while aligning with Microsoft’s continued investment in hardware-rooted trust.”

Popular tools designed to activate pirate copies of Windows have used KMS-based methods for quite some time. In 2025, Microsoft stopped the workaround employed in the so-called “KMS38” activation method, although “pure” KMS activation should still work as expected.

The elusive Massgrave collective provides open-source tools for “unofficial” Windows activation procedures, including an Online KMS method that needs to “phone home” to a fake KMS server every six months. TPM-based attestation might very much be the end of Online KMS-based piracy, although we will have to wait and see how the story actually ends. Massgrave recently introduced the new TSforge Activation method, which can allegedly bypass Microsoft’s entire DRM architecture for software product activation.