Back in March 2024, we reported how British AI startup Literal Labs was working to make GPU-based training obsolete with its Tseltin Machine, a machine learning model that uses logic-based learning to classify data.
It operates through Tsetlin automata, which establish logical connections between features in input data and classification rules. Based on whether decisions are correct or incorrect, the machine adjusts these connections using rewards or penalties.
Developed by Soviet mathematician Mikhail Tsetlin in the 1960s, this approach contrasts with neural networks by focusing on learning automata, rather than modeling biological neurons, to perform tasks like classification and pattern recognition.
Energy-efficient design
Now, Literal Labs, backed by Arm, has developed a model using Tsetlin Machines that despite its compact size of just 7.29KB, delivers high accuracy and dramatically improves anomaly detection tasks for edge AI and IoT deployments.
The model was benchmarked by Literal Labs using the MLPerf Inference: Tiny suite and tested on a $30 NUCLEO-H7A3ZI-Q development board, which features a 280MHz ARM Cortex-M7 processor and doesn’t include an AI accelerator. The results show Literal Labs’ model achieves inference speeds that are 54 times faster than traditional neural networks while consuming 52 times less energy.
Compared to the best-performing models in the industry, Literal Labs’ model demonstrates both latency improvements and an energy-efficient design, making it suitable for low-power devices like sensors. Its performance makes it viable for applications in industrial IoT, predictive maintenance, and health diagnostics, where detecting anomalies quickly and accurately is crucial.
The use of such a compact and low-energy model could help scale AI deployment across various sectors, reducing costs and increasing accessibility to AI technology.
Literal Labs says, “Smaller models are particularly advantageous in such deployments as they require less memory and processing power, allowing them to run on more affordable, lower-specification hardware. This not only reduces costs but also broadens the range of devices capable of supporting advanced AI functionality, making it feasible to deploy AI solutions at scale in resource-constrained settings.”
A new variant of the infamous ClearFake (AKA ClickFix) malware has been detected in the wild, and has already managed to compromise thousands of WordPress websites.
Researchers from GoDaddy claim to have spotted a variant of this campaign, which installs malicious plugins to sites on the website builder. The threat actors would use the credentials stolen elsewhere (or bought on the black market) to log into the website’s WordPress admin account, and install a seemingly benign plugin.
The victims are then enticed to download an update, which is just a piece of malware that steals sensitive data, or does something else but equally sinister.
Thousands of compromised websites
In turn, the plugin displays the various popups, requesting the victims do different actions (all of which lead to the installation of infostealers).
The entire process is automated, GoDaddy is saying, and so far more than 6,000 WordPress websites have fallen prey.
"These seemingly legitimate plugins are designed to appear harmless to website administrators but contain embedded malicious scripts that deliver fake browser update prompts to end-users,” the researchers are saying. The plugins are “seemingly legitimate” as they carry household names in the WordPress world, such as Wordfense Security, or LiteSpeed Cache.
Here is the full list of the plugins spotted so far:
LiteSpeed Cache Classic
MonsterInsights Classic
Wordfence Security Classic
Search Rank Enhancer
SEO Booster Pro
Google SEO Enhancer
Rank Booster Pro
Admin Bar Customizer
Advanced User Manager
Advanced Widget Manage
Content Blocker
Universal Popup Plugin
ClearFake is a type of malware attack we’ve all seen in the past - a website is compromised and used to display a fake popup notification. This notification usually mimics an antivirus warning, or a browser notification, and informs the user that their computer is either infected with a virus, or outdated and therefore unable to display the desired website.
The OnePlus 13 is set to be one of the final flagship phones of 2024, though it will probably only be a 2024 release in China, with the rest of the world likely to get it in 2025.
Either way though, it could be worth the wait, as between leaks and teasers we have a good idea of what to expect, and that includes an extremely powerful and efficient chipset, a huge battery, and a striking design.
Below then, you’ll find everything we’ve heard about the OnePlus 13 so far, including its confirmed announcement date and design, and leaked specs and features.
Cut to the chase
What is it? The next OnePlus flagship
When is it out? October 31 in China, maybe early next year globally
How much will it cost? Likely upwards of $799.99 / £849 (around AU$1,200)
A rumored price increase
The OnePlus 12(Image credit: Philip Berne / Future)
Landing in China on October 31
Could cost more than the OnePlus 12
The OnePlus 13 will be unveiled on October 31 at an event in China. The company itself has confirmed this, with the event kicking off at 1am PT / 4am ET / 8am GMT / 7pm AEDT.
This event will most likely just be the Chinese launch of the phone though, with a global launch probably happening in early 2025 based on past form.
The OnePlus 13 might also cost more than the OnePlus 12, with one leak suggesting a 16GB model with 512GB of storage will cost ¥5,299 in China, which is up from ¥4,799 for the equivalent model of the OnePlus 12. That’s a difference of around $70 / £55 / AU$105.
For reference, the OnePlus 12 cost $799.99 / £849 (around AU$1,200) at launch for a 12GB / 256GB model, and rose to $899.99 / £999 (roughly AU$1,350) for a 16GB / 512GB version. So the OnePlus 13 might have a higher price than that.
Oh, and don’t count on being able to buy it in Australia, because the current model isn’t available there.
Can you trust these rumors?
We've only heard one price leak so far so we'd take it with a pinch of salt. But it's believable that the OnePlus 13 would cost more than the OnePlus 12, especially as the Snapdragon chipset we're expecting it to use reportedly costs more than its predecessor.
A familiar design and a new screen technology
The OnePlus 12(Image credit: Philip Berne / Future)
A familiar design in new colors
The screen has an interesting refresh rate technology
The phone will be available in White Dawn (white), Blue Moment (blue), and Obsidian Secret (black) shades, with the blue one apparently having a skin-like feel and the black one having an “ebony wood grain” textured finish.
The design of the back otherwise looks somewhat similar to the OnePlus 12, but the circular camera block doesn’t extend out to the edge here, and instead there’s a silver line running across the rear.
The OnePlus 13 in three colors(Image credit: OnePlus)
We haven’t yet had a complete official look at the front of the phone, but you can see glimpses of it in a OnePlus 13 teaser video, showing what appears to be a flat display with a punch-hole camera in the top center.
We’ve also seen actual photos of the OnePlus 13, from where it was spotted at a Chinese esports event, as you can see below.
Images of the OnePlus 13 at an esports event(Image credit: Weibo)
Still, it shows a seemingly flat screen with a punch-hole camera, similar to other teasers above.
A possible image of the OnePlus 13(Image credit: OnePlus)
In any case, leaks suggest the OnePlus 13 might have a 6.82-inch display, which is the same as the OnePlus 12. An early leak also suggested the OnePlus 13 would have a curved screen, but based on more recent leaks and teasers that doesn’t appear to be the case, unless it’s a very subtle curve.
The company has also teased on Weibo that its screen will be able to deliver different refresh rates in different sections, so for example static text could be at 10Hz while a video playing elsewhere on the screen could be 120Hz. This is something we haven’t seen from other Android phones, and it might allow for improved battery life.
Most of the design has now been confirmed, and the leaked screen specs are all in line with what we'd expect. The one real question mark here is whether the OnePlus 13 will have an ultrasonic fingerprint sensor, but it sounds believable.
A mix of old and new cameras
The OnePlus 12(Image credit: Philip Berne / Future)
It would be a shame not see any upgrades there, but in an earlier leak the same source claimed OnePlus was working on the OnePlus 13’s periscope camera, so that at least might differ from the 64MP 3x zoom of its predecessor.
As for exactly how it might differ, well, both that source (Digital Chat Station, via Android Authority) and leaker Yogesh Brar have claimed that the OnePlus 13 might have three 50MP cameras. In which case that would also mean changes to the ultra-wide, as that’s 48MP on the OnePlus 12.
Can you trust these rumors?
The cameras are one of the things we're least sure about, as we haven't heard all that much about them. But the leaks we have heard generally all point in the same direction, so the specs above have a good chance of being accurate.
An efficient chipset and a huge battery
The OnePlus 12(Image credit: Philip Berne / Future)
That should help improve battery life, as should the ability to deliver different refresh rates to different parts of the screen, as mentioned further up in this article. And that’s on top of the OnePlus 13 reportedly having a massive 6,000mAh battery, so it’s likely to excel for stamina.
Leaks also suggest the OnePlus 13 will charge quickly, at up to 100W wired and 50W wireless – though based on past form the US might get slightly lower 80W wired charging speeds.
As for other specs, we’ve heard from leaker Digital Chat Station that the OnePlus 13 could come with up to 24GB of RAM, though the OnePlus 12 also did, but only in China, so there’s every chance international models will once again top out at 16GB.
Can you trust these rumors?
There are no red flags among the leaked battery or chipset details, and in fact we're almost certain the OnePlus 13 will use the successor to the Snapdragon 8 Gen 3.
Qualcomm’s Snapdragon Elite chips have been the star of many AI PCs released to market in 2024, greatly enhancing performance and battery life compared to their previous gen iterations. However, the manufacturer’s recent endeavor to improve on one of its negative points hasn’t panned out.
Qualcomm has canceled its dev kit, a Snapdragon mini-PC powered by Windows on Arms. Originally, it was planned for a June 2024 release window but missed that entirely. Now, as part of an official announcement, Qualcomm has stated that the kit has been canceled because it “has not met our usual standards of excellence.”
Interestingly enough, The Verge spotted that multiple developers had already received the mini-PC including YouTuber Jeff Geerling, who reviewed the product and performed a detailed hardware teardown of it. Despite Qualcomm not providing any specific reasons for why the PCs had been canceled, the conclusions from Geerling’s review are that despite it matching the Apple M3 Pro’s performance, it lacks Linux support and had restrictions around reselling the device.
There’s another theory for the cancellation as well, which comes down to the HDMI port — or lack thereof. Though the shipped mini-PCs all came with the chips for internal DisplayPort to HDMI conversion, the actual port was missing for some reason.
According to Richard Campbell, founder of DEVIntersection, the HDMI port could have been the cause of the massive delays if it failed FCC compliance testing. This seems to be supported by the fact that Qualcomm emailed anyone who ordered the PC in September that they would be sent a USB-C to HDMI dongle with their dev kit.
But what does this mean for the consumer?
One of the most well-known drawbacks to anything Qualcomm is that Windows on Arms still has compatibility issues with some Windows programs, tools, and apps. The manufacturer has improved this by leaps and bounds to the point that the average user wouldn’t know the difference, but for gamers and others using specialized programs, those compatibility problems can be quite difficult to parse.
In comes the Qualcomm mini-PC dev kit, which would have been the perfect tool for developers to port their apps to Windows on Arms. This potentially could have introduced a large amount of apps to the OS that otherwise would have never seen the light of day on Arm chips. This is especially important for the consumers who have been left in the dark due to the lack of app support for tools that they need themselves, limiting Qualcomm’s sales of its AI PCs to them in turn.
And with fellow industry rivals Intel and AMD teaming up to form the “x86 Ecosystem Advisory Group,” in response to Qualcomm and other competitors, the latter must figure out either how to fix the issues with the dev kits or figure out another way to resolve these app compatibility issues if it wants to see Arms continue to thrive.
Ferrari has lifted the lid on the most powerful road-going supercar it has ever built, cramming the carbon fibre-laden F80 with a three-liter twin-turbo six cylinder engine that develops 900hp on its own.
The combustion engine is then paired up with three electric motors (all developed in house) that throw an additional 300hp into the mix. That’s 1200 horses in total from a propulsion system that is essentially derived from its 499P Le Mans race car, but is equally as happy popping to the shops for a pint of milk.
Far from an attempt to reduce Ferrari’s overall carbon emissions, this is instead an exercise in extracting maximum performance. The diminutive 800V/2.3kWh battery that is stashed behind the seats, for example, uses F1 technology to ensure it can charge and discharge in the quickest and most efficient way possible.
That means customers won’t be able to drive the F80 in an all-electric mode, but will instead be able to rocket from a standstill to 62mph in 2.15 seconds, reach 124mph in just 5.75 seconds and max out at 217mph. Mind-melting stuff.
(Image credit: Ferrari)
The F80 is engineered for raw performance, using a carbon-fiber chassis that is as narrow as engineers could get away with to ensure less aerodynamic drag. Inside, there’s just about enough room for the driver and one very small passenger.
Ferrari calls it a "1+" vehicle, seeing as the second seat is slightly staggered to allow the controls to envelope the driver. To highlight that this isn't some romantic grand tourer, the driver’s seat is bright red, while the poor passenger has to make do with black... and race harnesses.
There’s so much tech trickery going on underneath the spaceship-style body and butterfly wing doors that engineers from Maranello could easily fill an entire Netflix documentary series with content. The motor-controlled active suspension system, for example, negates the need for anti-roll bars and allows the vehicle to transform from hunkered down track monster to a pliant road ride with ease.
Ferrari has harnessed 3D printing technology to create the suspension wishbones, and also used copious amounts of carbon fiber for its five-spoke alloys wheels and has also introduced some serious software trickery to squeeze the most out of race circuit lap times.
Boost Optimization, which race fans can access in the two high performance driving modes (Performance and Qualify), maps warm-up laps of a favored circuit and then automatically delivers additional horsepower in the section that it deems need it most, such as when accelerating along a straight.
A technological powerhouse
(Image credit: Ferrari)
The Ferrari F80 won’t actually go into production until 2025 – from then, the 799 lucky customers that are selected to receive one will have to wait until 2027 to experience it. Handily that's Ferrari’s 80th anniversary, hence the name.
It will cost around €3.6million, which is over $4million and almost AU$6million, by far the most expensive Ferrari road car ever built and up there with some of most expensive production super cars of all time.
But it moves the game on for the Italian automaker, not just in the raw figures when compared to the LaFerrari it usurps but also in the active aerodynamic technology, track-focussed features and extensive use of e-motors in everything from the turbochargers to the suspension system.
The proof is in the pudding, as the marque claims the F80 posted a new lap record of its Fiorano test track of 1min 15.3secs – that’s 4.4 seconds faster than the LaFerrari.
This might sound like marginal gains, but the numbers are huge in the cutthroat world of low-volume but hideously expensive super cars.
The Sun’s corona, its outermost atmospheric layer, holds the key to understanding solar activity, including phenomena like solar flares and space weather events. For decades, scientists have grappled with the challenge of measuring the Sun’s coronal magnetic field, as this field drives much of the energy that leads to solar eruptions.
Now, in a groundbreaking achievement, Professor Tian Hui's research team from Peking University, in collaboration with international experts, has made the first conventional measurements of the global coronal magnetic field. Their findings, published in the journal Science (Volume 386, No 6717), offer new insights into the Sun’s magnetic activity over a period of eight months.
The Sun’s magnetic field is responsible for storing and releasing energy, which heats the plasma in the corona and triggers solar eruptions. These eruptions, in turn, can have significant impacts on space weather, potentially affecting satellite operations, GPS systems, and even human spaceflight. However, due to the relatively weak nature of the coronal magnetic field compared to the magnetic field on the Sun’s surface (the photosphere), measuring this field has proven to be a significant challenge.
The importance of coronal magnetic field measurements
As the Sun rotates, there are variations in the magnetic fields and the ability to regularly monitor the Sun’s coronal magnetic field will improve our understanding of solar eruptions and help protect high-tech systems on Earth and in space.
Over the years, there have been routine measurements of the photospheric magnetic field, but the coronal field has remained elusive. This limitation has impeded scientists' ability to fully understand the three-dimensional magnetic field structure and the dynamic processes that occur in the Sun’s atmosphere.
In 2020, Tian Hui's team developed a method called “two-dimensional coronal shocks,” which enabled the first measurements of the global distribution of the coronal magnetic field. This was a major milestone, marking a crucial step towards the goal of routine coronal magnetic field measurements.
More recently, Tian's team refined this method further, allowing them to track magnetohydrodynamic shear waves in the corona with greater precision. This made it possible to diagnose the distribution of coronal density and, as a result, determine both the strength and direction of the magnetic field.
Using the Upgraded Coronal Multi-Channel Polarimeter (UCoMP), the research team conducted detailed observations of the Sun’s corona from February to October 2022. Over this eight-month period, they collected 114 magnetograms, or magnetic field images, which allowed them to observe the evolution of the coronal magnetic field at different altitudes and latitudes across multiple solar rotations. The magnetic field strength measured between 1.05 and 1.60 solar radii and varied from less than 1 gauss to around 20 gauss.
With these measurements, they were able to create a global map of the magnetic field intensity in the Sun’s corona. This map revealed how the magnetic field evolves over time and across different regions of the Sun.
When compared to the predictions of the most advanced global coronal models—such as the one developed by Predictive Science, a U.S.-based company—the team found that their observational data closely matched the model’s predictions in mid- and low-latitude regions. However, they noted larger discrepancies in high-latitude regions and active areas of the Sun.
These findings are critical for improving current models of the Sun’s magnetic activity and understanding the dynamics of solar eruptions. As lead author Yang Zihao explains, the team's observations provide a key foundation for refining and optimizing coronal models, which could eventually lead to more accurate predictions of solar eruptions and their potential impact on Earth’s space environment.
This study marks a shift in solar physics, as the field enters a new era of routine coronal magnetic field measurements.
According to Tian Hui, this achievement is just the beginning. While their current methods allow for the measurement of the magnetic field at the edge of the solar disk, the next goal is to develop techniques that will enable a complete measurement of the entire coronal magnetic field, including the solar disk itself. This will require the integration of other measurement methods and tools, but it represents a critical objective for the solar physics community in the coming decades.
