Nearly all Chinese keyboard apps have encryption flaws, exposing millions of users to keylogging

In brief: Anyone who types in Chinese using cloud-based apps from Baidu, Honor, iFlytek, Oppo, Samsung, Tencent, Vivo, and Xiaomi should update their software immediately. The Huawei service appears to be safe, but security flaws in the other apps, most of which recently received patches, could invite snooping, potentially impacting up to a billion users.

Researchers recently discovered severe encryption flaws in cloud-based pinyin input software from eight companies that could allow eavesdropping. Although there is no evidence that the vulnerabilities are actively being exploited, earlier incidents make this a potentially serious issue.

Chinese writing incorporates thousands of unique characters that could never fit on conventional keyboards, so typing in the language requires alternate input methods (IMEs). All of the vulnerable cloud tools employed pinyin systems, in which users type phonetic pronunciations using Roman letters and then pick from a selection of corresponding symbols. Operating system vendors and third-party developers have provided Chinese IMEs with wholly on-device processing for decades, but cloud services can determine the correct characters more accurately.

Normally, any internet-based typing utility comes with inherent risk, but the companies offering cloud-based pinyin apps guarantee user privacy through encryption. Researchers from the University of Toronto tested the security of apps from nine companies: Baidu, Honor, Huawei, iFlytek, OPPO, Samsung, Tencent, Vivo, and Xiaomi, and successfully read keystrokes from all of them except the Huawei tool, potentially exposing every input to eavesdroppers. Some of the vulnerabilities can leak data to completely passive snoopers.

Notably, the researchers found no flaws in iOS apps because Apple automatically sandboxes the platform's keyboard apps. Allowing iPhone keyboard apps to access and transmit data requires explicit user permission. Meanwhile, equivalent Android and Windows tools were deemed far less secure. Android users can choose whether keyboards connect to the internet, but the researchers found that the corresponding controls could be too difficult for some users to find.

After the researchers warned all nine vendors, most released updates to fix the problems, but encryption flaws remain in Baidu's apps, Honor's keyboard, and Tencent's QQ Pinyin service. Furthermore, the researchers listed dozens of similar apps that they couldn't test but might suffer from the same issues.

The researchers expressed alarm partly due to prior episodes involving government surveillance. The report notes that the Five Eyes – an intelligence-sharing alliance between the US, UK, Canada, Australia, and New Zealand – had previously used similar vulnerabilities in Chinese apps to spy on their users.

Cybercriminals are using developing nations as test beds for ransomware attacks

In brief: With ransomware becoming so lucrative and widespread, hackers have started deploying new malware strains in developing and emerging nations as a way of testing them before attacking businesses in richer countries. Organizations in Africa, Latin America, and Asia are hit first as they tend to have weaker security and draw less attention. The hackers then attack high-profile targets in North America and Europe.

Attacks have been observed on a bank in Senegal, a financial services company in Chile, a tax firm in Colombia, and a government economic agency in Argentina using strains of malware that were later used in Europe and North America, writes cybersecurity firm Performanta (via Ars Technica).

One of these ransomware strains is Medusa, a variant that gets its name from being able to turn files "into stone" by stealing and encrypting data. It was first used against businesses in South Africa, Senegal, and Tonga in 2023. It was later used in 99 breaches in the US, UK, Canada, Italy, and France.

Medusa victims would see a file with the subject line !!!READ_ME_MEDUSA!!!.txt. instructing users to start negotiations with the ransomware gang on the dark web. Failure to do so would result in the stolen data being published online.

Nadir Izrael, chief technology officer at cyber security group Armis, said that when attackers were discussing a new vulnerability, named CVE-2024-29201, earlier this year, they "specifically targeted a few [exposed servers] in third world countries to test out how reliable the exploit was." The gang's attacks were restricted to South East Asia before becoming more widespread.

Teresa Walsh, chief intelligence officer at global cyber threat intelligence body FS-ISAC, said some gangs perfect their ransomware techniques in poorer countries, such as Brazil, against less well-protected companies before exporting their attacks to richer nations that speak the same language, like Portugal.

Sherrod DeGrippo, director of threat intelligence strategy at Microsoft, has a different take. She believes the increase in attacks on organizations in developing countries is due to ransomware gangs selling their product to less-sophisticated hackers in poorer nations. These attackers often do not understand how the malware works, so they only stage their attacks against less well-guarded targets.

In other ransomware news, a member of the notorious LockBit ransomware group was sentenced to four years in jail last month for infecting over 1,000 systems.

Generative AI could soon decimate the call center industry, says CEO

A hot potato: It's no secret that certain types of jobs are more threatened by artificial intelligence than others. Call center workers fall into this category, and while we've already seen a few companies replace phone-based support staff with generative AI, there are warnings that the entire industry could be comprised mostly of chatbots in as soon as a year.

The grim prediction comes from K Krithivasan, head of Indian IT giant Tata Consultancy Services (TCS). The second-largest company in India by market cap, it has more than 616,000 employees worldwide.

Speaking to the Financial Times, Krithivasan said AI will result in a "minimal" need for call centers. The CEO added that while "we have not seen any job reduction" so far, that will change as multinational clients adopt generative AI. The technology is expected to have a massive impact on the customer help center industry, which, according to a Gartner report in 2022, employs about 17 million people.

"In an ideal phase, if you ask me, there should be very minimal incoming call centres having incoming calls at all," Krithivasan told the FT. "We are in a situation where the technology should be able to predict a call coming and then proactively address the customer's pain point."

The prospect of a chatbot being able to fulfil all of a customer's requests over the phone with ease might sound like a long way off, but Krithivasan believes they will be able to seamlessly replace humans in "maybe a year or so down the line."

We've already seen companies oust call center staff in favor of AI. In July last year, a CEO laid off 90% of his support team, boasting that the move dropped first response and resolution times while reducing customer support costs by around 85%. He later warned that the technology is "100%" going to kill copy-paste jobs.

Krithivasan did add that the immediate impact of generative AI was overblown and that people were still overestimating the benefits. He also argued that the technology wouldn't lead to mass job losses as the world would need more people "in terms of technology talent," but then most CEOs like to push the narrative about gen AI assisting people rather than replacing them, and how its use will lead to new types of jobs.

There have been several surveys estimating the impact generative AI will have on jobs. One said it is expected to replace two million jobs in the US by 2030, while the IMF believes 40% of jobs worldwide will be affected in some way by the technology. Possibly the most worrying survey revealed that almost half of all managers aim to replace workers with AI and could use it to lower wages.

Masthead: Wikkimedia commons

Nintendo DMCA lawyers shut down everything Mario on Garry's Mod

Why it matters: It's almost hard to believe that Garry's Mod has been around for 20 years. What started as a simple mod built on Valve's Source Engine in 2004 grew into a standalone sandbox "game" by 2006, where users could experiment with 3D models and the physics engine. Eventually, Garry's Mod allowed user-created games and content, evolving it into a platform for creation.

On Wednesday, Garry's Mod developer Facepunch announced that it would pull all Nintendo-related content from Steam Workshop. The takedown is in response to Nintendo serving DMCA notices to several content creators and Facepunch. The studio said the process would take some time because it has to shuffle through two decades of uploads. It asked the community for help.

"This is an ongoing process, as we have 20 years of uploads to go through," Facepunch said on Steam. "If you want to help us by deleting your Nintendo-related uploads and never uploading them again, that would help us a lot."

[PSA] False flaggers on the Garry's Mod Workshop.

There is someone, or a group of trolls going around the Gmod workshop filing false DMCA claims and shutting down addons, over 10 Gmod addon makers, myself included have been attacked, with over a dozen of our addons gone. pic.twitter.com/ZwVVhQjQlu

– Brewster T. Koopa (@brewsterkoopa) February 11, 2024

The DMCA notices started with individual Garry's Mod contributors, spurring speculation that the requests were bogus. A content creator who goes by Brewster T. Koopa on X called the notices "false flags" from trolls. He pointed out that the takedowns couldn't have come from Nintendo because the add-ons being scrubbed date back to 2005 and that the lawyers would have gone after Valve, not individual creators.

When add-ons began coming down, Koopa emailed Facepunch to inform it that the notices were fake. Unfortunately, Facepunch had already reached out to Nintendo, which verified that the takedowns were legit.

If the initial threats were bogus, Facepunch shot itself in the foot by contacting Nintendo. The highly litigious company is not beyond taking advantage of a presented DMCA situation. Unfortunately, whether the notices were fake or real is a moot point; the result is the same – the content must come down.

"This is not a mistake, the takedowns came from Nintendo," Facepunch said on Wednesday. "Honestly, this is fair enough. This is Nintendo's content, and what they allow and don't allow is up to them. They don't want you playing with that stuff in Garry's Mod – that's their decision, we have to respect that and take down as much as we can."

I have been assured that the takedowns have been verified by Nintendo as legit, so this will now continue as planned. Sorry. ð«¸ðÂÂÂ�-'ï¸Â

– garry (@garrynewman) April 25, 2024

Yesterday's announcement of the removals prompted even more emails, direct messages, and forum threads from the community crying that it's just trolls, some providing "proof." The response was significant enough that Facepunch founder Garry Newman tweeted in the wee hours of Thursday morning that the studio was launching an investigation.

Unfortunately, that investigation ended about nine hours later when Nintendo reiterated to Facepunch that it was serious about the takedowns.

"I have been assured that the takedowns have been verified by Nintendo as legit, so this will now continue as planned. Sorry," Newman tweeted.

BlizzGone: Blizzard cancels 2024 convention but promises an eventual return

Bottom line: Blizzard Entertainment has canceled the 2024 installment of BlizzCon, the annual gaming convention usually held near the end of the year in Anaheim, California. The company didn't give a concrete reason as to why this year's event is now off, simply noting that they decided against it after careful consideration over the last year. That said, officials look forward to bringing BlizzCon back in future years.

The decision could have something to do with the company now being under Microsoft's umbrella. As such, we can probably expect to learn more about upcoming Blizzard projects like World of Warcraft: The War Within and the first Diablo IV expansion, Vessel of Hatred, during Microsoft's annual Xbox showcase in June.

Blizzard also mentioned plans for other trade shows like Gamescom and additional in-person gathers, like upcoming Overwatch Champions Series' stops at Dreamhack Dallas and Dreamhack Stockholm. There will even be multiple global, in-person events to celebrate the 30th anniversary of Warcraft (the first game in the franchise, Warcraft: Orcs and Humans, launched on November 15, 1994).

Blizzard said that while these other events will be distinct from BlizzCon, they are working to harness their creativity and imagination to ensure they carry the same spirit of togetherness and celebration.

In-person gaming and tech conventions have struggled in recent years, and BlizzCon is no exception. The gathering was canceled in 2020 due to Covid and was held as an online only event the following year. Blizzard also canceled the 2022 convention due to legal issues at the time. Things got back on track in 2023, but it was short lived.

E3 experienced similar pitfalls following the pandemic. Last December, however, organizers threw in the towel and canceled the event for good. Fortunately, others have stepped in so the loss shouldn't be as stinging as it might otherwise have been.

Some also thought similar conferences like CES were in jeopardy, but organizers managed to host an abbreviated event in 2022 before returning in full force in 2023.

Astronauts aboard the ISS craft miles of technically superior optical fiber

Space manufacturing: Optical fibers are used in a variety of industries where high-bandwidth data transmission over long distances is paramount. It's an impressive bit of engineering but as recent testing on the International Space Station has proven, fiber manufactured in space is far superior to what we can make here on Earth.

Astronauts used ZBLAN preform, which is a glass alloy compromised of barium, zirconium, lanthanum, aluminum fluorides, and sodium, in their tests. These glass rods are used in runs (called draws) in which they are stretched into incredibly thin fibers.

When ZBLAN is drawn into fibers on Earth, gravity causes crystals to form that scatter light and make the fiber brittle. Crystals form slower in microgravity, however, and can even be avoided entirely if the draws are cooled down fast enough.

ZBLAN-based optical fibers allow light to travel further than traditional silica-based fibers, and have more than 10 times the data capacity.

Preliminary results from NASA's testing look promising. Astronauts were able to draw more than seven miles (11.9 km) of fiber over a month-long production run on the ISS. NASA said eight of its draws produced more than 2,200 feet (700 meters) of fiber. In another test, they were able to draw more than 3,700 feet (1,141 meters) in a single day.

The space-drawn, ZBLAN-based fibers are expected to return to Earth soon for quality analysis. Rob Loughan, CEO of Flawless Photonics (the company that made the space manufacturing equipment), told The Register that samples will eventually be sold to companies interested in experimenting with space-drawn, ZBLAN-based fiber.

According to Loughan, they have already secured several orders for space-made fiber at $1,00 per meter. One client even purchased $1,000 worth, or about 100 meters.

Flawless is also working to improve the ZBLAN purification process, and eventually aims to make its own preforms in space to bring home and see how they respond to being pulled under Earth's gravity. It'd also be interesting to see how ZBLAN made in space compares to Earth-made ZBLAN when drawn on the ISS.

Lenovo and Micron are first to announce a laptop using LPCAMM2 memory

Forward-looking: Lenovo is on track to become the first PC maker to ship a laptop running LPCAMM2, the new memory standard set to replace SODIMM. CAMM, short for Compression Attached Memory Module, was created by Tom Schnell and introduced by Dell in its Precision 7670 workstation laptop in 2022. Rather than hang on to the proprietary tech, Dell agreed to let JEDEC – the trade organization and standardization body that oversees memory standards in the computer industry – use it as the basis for a new memory standard.

JEDEC finalized and published the CAMM2 standard in December 2023, paving the way for Micron to announce its first LPCAMM2 modules at CES the following month. Now, Lenovo is poised to become the first to ship a laptop running the newfangled memory.

The upcoming AI-ready ThinkPad P1 Gen 7 can be configured with up to 64 GB of LPDDR5x LPCAMM2 memory. According to Lenovo, its LPCAMM2 solution consumes up to 61 percent less active power and affords a space-saving advantage of 64 percent compared to traditional DDR5 SODIMM.

Elsewhere, the ThinkPad P1 Gen 7 features up to a 16-inch OLED touch screen with a 16:10 aspect ratio, and can be combined with up to an Intel Core Ultra 9 185H CPU, and as much as 8 TB of PCIe M.2 2280 SSD space. Graphics options include integrated Intel Arc on the low end or an Nvidia GeForce RTX 4070 for serious gamers.

Other noteworthy amenities include up to Wi-Fi 7, Bluetooth 5.3, and a plethora of connectivity ports (two Thunderbolt 4, one USB-C (10Gbps), one USB-A (5Gbps), an SD Express 7.0 card reader, an HDMI 2.1 port, and a standard audio jack). The laptop also comes with a Kensington Nano lock slot, a webcam privacy shutter, and a biometric-enabled power button.

Lenovo made no mention of battery life, but did say the 90Whr unit inside the laptop is customer replaceable.

The Lenovo ThinkPad P1 Gen 7 launches in May 2024 and is expected to start at $2,619. We'll keep our eyes peeled for the first batch of reviews to see how the new memory compares to traditional SODIMM.

Upgraded 4K Chromecast with Google TV set to launch soon with improved hardware

Rumor mill: Google is reportedly gearing up to unveil its second-generation Chromecast with Google TV 4K four years after launching the first-gen model. The company also launched a more affordable HD variant in 2022, but this promises to be a full-fat sequel to the original model, complete with 4K streaming capabilities.

The report comes from 9to5Google, which claims that the Android-powered dongle will ship with a new remote, but keep its $49.99 price-tag, possibly making it a real competitor against other similarly-priced streaming devices from Amazon and Roku.

There's no word on what to expect from the new device in terms of its hardware, but Google will likely implement some changes to make it more powerful than its predecessor. For reference, the 2020 model is powered by an Amlogic S905X3 processor, which has a quad-core Arm Cortex-A55 CPU clocked at 1.9 GHz. It also includes 2 GB of RAM and 8 GB of internal storage.

Chromecast users looking to upgrade from the existing model will be hoping that the new variant will bring a faster processor, more memory and additional storage. Online speculations suggest that it could come with the newer Amlogic S905X5 chipset that is expected to be faster and more power efficient than the older processor.

The S905X5 is built on a 6nm manufacturing process, which represents a generational change over the 12nm node of the X3 and X4. It also includes a new CPU with Cortex-A510 Armv9 cores and a Mali-G310 V5 GPU. Along with these hardware improvements, the S905X5 supports H.266 decoding. New Android TV boxes powered by the new chip have already been announced, although they are yet to hit the market.

As for the rumored new remote, a video found within Android TV 14 beta last year seemingly revealed that it could have a redesigned button layout that includes four large round buttons on the left side and two large round buttons plus a third large oblong button on the right. According to Android expert and former XDA editor-in-chief Mishaal Rahman, there will also apparently be a customizable "Magic button" that could let users choose whether pressing the button will "open a favorite app" or "view and switch inputs for TV or other devices."

Tech giants are spending more on AI amid slow returns, and that's spooking investors

In brief: Tech companies continue to pour billions of dollars into artificial intelligence-related areas, from hardware and software to general investments and acquisitions. But reaping the financial rewards is taking longer than expected. Meta's announcement that it is spending more on AI while profitability remains a long way off has spooked investors, and other companies are feeling the same pressure.

Meta said that it would be spending billions of dollars more on its AI efforts, raising expenditure forecasts in this area from $30 - 37 billion to $34 - 40 billion for the year. It cited heavy investment in AI infrastructure such as data centers, chip designs, and research and development for the increase. Meta also predicted that revenue for the current quarter would be lower than expected.

During an earnings call with investors, Meta CEO Mark Zuckerberg said that profitability from generative AI will take years, but asked for patience. "Smart investors see that the product is scaling and that there is a clear monetizable opportunity there even before the revenue materializes," he said. Zuckerberg also pointed to the energy costs of generative AI for Meta's higher spending.

Zuckerberg has experience when it comes to assuring investors that new tech costing the company a fortune will eventually pay off. Meta's Reality Labs, the VR/AR business responsible for its metaverse ambitions, has hemorrhaged more than $42 billion since the end of 2020, but Zuckerberg continues to insist that the metaverse will make billions or even trillions of dollars after 2030.

Meta's not the only one spending more on AI. New Street Research analysts predict Google parent Alphabet will see its full-year capital expenses increase to about $45.9 billion, up from previous estimates of $42.7 billion, partly as a result on spending in this field, writes Reuters.

Microsoft has also spent billions in this area, including the more than $10 billion investing in OpenAI. Like Alphabet, Microsoft said earlier this year that it expects its AI-related costs to keep rising.

Despite Meta's income growing 27% YoY in the first quarter to $36.5 billion, beating Wall Street's predictions, and profit more than doubling to $12.4 billion, the high spending and lower-than-expected revenue for the current quarter saw the company's stock crash 16%. Microsoft is down 2%, Alphabet fell 3%, and Nvidia was down 1.4% as a result.

Meta AI assistant was pushed out across the company's suite of apps last week. It's currently free to use, but Meta believes there are plenty of ways to monetize the feature, such as enabling people to pay to use bigger AI models and access more compute. Zuckerberg also believes AI will improve app engagement, leading to more ad money. Whether that will calm nervous investors looking for near-term gains remains to be seen.

Seagate says their new Mozaic 3+ HAMR disk drives can last longer than conventional HDDs

Burning Drives: Heat-assisted magnetic recording (HAMR) is the first major hard disk drive technology advancement to arrive in a very long time. Seagate beat everyone by rushing its Mozaic 3+ platform to the market, and the company is now trying to reassure customers and industry partners about the reliability of the new storage solution.

The Mozaic 3+ hard drive platform is as reliable, or even more reliable, as conventional HDD technology. Seagate recently introduced the new hard disk drives based on its own implementation of the HAMR technology, with 30 TB HDDs available to customers and a very ambitious plan to push storage technology further with 50TB units coming in the not so distant future.

The US corporation spent more than $2 billion to develop Mozaic 3+ and the first commercial HAMR drives, providing new levels of mass-capacity storage with areal densities of 3TB per disk and beyond. Seagate undertook a significant R&D effort to bring HAMR to the market, with Mozaic 3+ including what the company describes as the most complex "nanoscale recording technologies" and material science breakthroughs on the planet.

HAMR is a transformative technology, so much that Seagate felt the need to provide an official "narrative" about the new solution. Misconceptions about reliability, compatibility, and ease of integration into the existing storage infrastructure are just industry myths, the company said, while HAMR has proven to be an extremely reliable storage technology in the firm's own tests.

Seagate tested over 500,000 Mozaic 3+ hard drives, and worked with customers and industry partners to gather feedback over HAMR implementation. Mozaic 3+ drives can seamlessly blend into the current data center ecosystem, the company added, as they employ the same HDD standards and protocols used by existing units. The rated mean time between failure (MTBF) for the new drives is 2.5 million hours, which aligns to the most recent perpendicular magnetic recording (PMR) enterprise hard drives in the Exos line.

"Thousands" of Mozaic 3+ read/write heads have been tested to prove their durability, and they were able to "reliably" transfer data for more than 6,000 hours – or 3.2 petabytes per single head. This is more than 20 times the amount of data transferred on a typical nearline hard drive. Field stress tests have shown that Mozaic 3+ read/write heads can last seven or more years, exceeding most of the current PMR-based products (5 years).

Seagate is betting on the fact that the transition from conventional PMR to HAMR recording will be seamless, as no changes will be needed to adopt the new drives into storage infrastructures that are already compatible with the "most recent" PMR drives. Global demand for data storage solutions keeps growing, and prices are surging because of AI hype. Seagate wants its new HAMR hard disk drives to be welcomed as a significant and reliable new advancement in the storage technology market.

TSMC unveils A16 manufacturing process for 1.6nm chips in challenge to Intel's claims

What just happened? TSMC has challenged Intel's claims that it will overtake the Taiwanese company when it comes to making the world's fastest computing chips by unveiling its A16 (1.6nm) manufacturing process. TSMC says the massive demand from unnamed AI customers meant the A16 technology has been developed faster than expected.

A16, set to enter production in the second half of 2026, will use leading nanosheet transistors and what TSMC calls an "innovative" backside power rail solution to bring "revolutionary performance" to the wafer level. While traditional chip designs involve power flowing from the top down, the backside power rail system delivers power from the bottom. The company writes that this will help address future AI requirements for hyperscale datacenters.

Intel says it also plans to integrate backside power rail technology into its 20A (2nm) and 18A (1.8nm) processes as early as 2025.

According to TSMC's roadmap, its 2nm N2 process node is on track for production in the second half of 2025, the successor to its 3nm N3E technology that is now in production. A16 will follow in 2026, combining TSMC's Super Power Rail architecture with its nanosheet transistors. It improves logic density by dedicating front-side routing resources to signals, which should make it ideal for HPC products. Compared to N2P, A16 will provide an 8-10% speed improvement, 15-20% power reduction, and up to 1.10x chip density improvement.

Kevin Zhang, TSMC's senior vice president of business development, said the company doesn't believe it will need to use ASML's new High NA EUV lithography tool machines for the A16 process.

Intel said last week that it will be the first to use ASML's machines, which cost $373 million each, to help develop its 14A (1.4nm) chips. Team Blue said in February that 14A would help it overtake TSMC in making the world's fastest chips. Samsung, meanwhile, is aiming to mass produce its 1.4nm chips by 2027.

Intel's foundry business lost $7 billion last year, something CEO Pat Gelsinger put down to past mistakes, including the decision to not use ASML's EUV machines. Intel later went back on its decision and invested in the EUV machines, which should help the foundry business' goal to break even in about three years.

TSMC emphasized the benefits the A16 process will bring to AI chip makers. Zhang said these firms "really want to optimize their designs to get every ounce of performance we have." Earlier this year, TSMC founder Morris Chang predicted that up to 10 new fabs would be required for AI chip manufacturing.

US TikTok ban could begin next year as Biden signs law, but legal battle looms

What just happened? After years of threats, a bill that could see TikTok banned from the US is now law. However, the ban would trigger no earlier than the beginning of next year, and a court battle could lengthen or even halt the process as a social network used by nearly half of Americans hangs in the balance.

US President Joe Biden has signed a bill into law that forces TikTok owner ByteDance to either sell the immensely popular social network or remove the app from Apple and Google app stores. The Beijing parent company has 270 days to comply, meaning the ban could enter effect next January at the earliest, with a possible three-month extension if a sale is in progress.

TikTok CEO Shou Chew responded by vowing to mount a legal challenge against the legislation, which could delay the issue further. In the meantime, he asked TikTok users to post testimonials explaining how important the platform is to them.

Our CEO Shou Chew's response to the TikTok ban: pic.twitter.com/l0RAPJMobK

– TikTok Policy (@TikTokPolicy) April 24, 2024

Around 170 million Americans – almost half the country's population – report having used TikTok, and it's the fastest-growing social media network in the US. Many users protesting the ban cited how integral it has become to their businesses and livelihoods.

A testament to the platform's reach is the Biden re-election campaign's intention to continue using TikTok as the president tries to force its sale or ban, setting the deadline after the November election. A campaign spokesperson said that the Biden campaign still wants to reach voters wherever they are.

However, US officials worry that TikTok harvests user data and shares it with ByteDance in China, accusing the companies of being under Chinese Communist Party control. Although TikTok transferred its US data to servers in Texas and claims that the employees responsible for it don't answer to ByteDance, anonymous employees recently told Fortune that the separation between the two companies is largely superficial.

Still, some argue that TikTok's data harvesting practices are no worse than those of Instagram or Twitter and that the law sets a double standard for the Chinese-owned platform. Apple co-founder Steve Wozniak called the ban "hypocrisy," comparing its practices with those of Facebook and Google.

Chew claimed that the constitution is on TikTok's side, likely referring to the First Amendment grounds upon which the company could base its defense. The strategy already has precedent, as a Montana judge blocked a statewide ban on the social network late last year, saying that it violated the Constitution in multiple ways, including the right to free speech.

Another concern is whether a ban from US app stores could truly keep Americans away from TikTok. US users could theoretically sideload the Android version of the app, and something similar could potentially occur on iOS if regulators force Apple to open its walled garden as the European Union has. It's also unclear how the law could stop Americans from accessing TikTok through web browsers and VPNs.

Nokia phone-maker HMD launches its first self-branded mobile devices

What just happened? HMD, the maker of current Nokia-branded phones, has released its first handsets featuring the company's own name. The devices are budget-friendly while also offering high levels of repairability. They're also environmentally friendly.

Finnish company HMD said in September that it would be establishing an original HMD brand of mobile devices that will exist alongside Nokia phones. Those devices have now been confirmed as the HMD Pulse, HMD Pulse Plus, HMD Pulse Pro, and HMD Vibe. The only one coming to the US is the HMD Vibe, which will cost a very wallet-friendly $149.

All three Pulse models share several similarities: they use the same Unisoc T606 chipset, have a 6.65" LCD screen, a 720p+ panel (20:9), and a 90Hz refresh rate. HMD says they can reach 600 nits and have 70% NTSC coverage.

All models come with Android 14 out of the box and the company promises 2 OS updates and 3 years of security patches. They feature 3.5mm headphone jacks, USB-C charging, 4G, Wi-Fi 5 (ac), Bluetooth 5.0, NFC, and GPS/Galileo. They're available in a range of snazzy colors, too.

The base Pulse model has a 13MP primary camera, a depth sensor, and an 8MP hole punch selfie cam on the front. It's IP52-rated for dust and water resistance, and there's an up to 59-hour, 5,000mAh battery that can be replaced within about five minutes using basic tools (i.e., it's not swappable). The battery is rated for 800 full charge cycles and uses 10W charging. There's also a side fingerprint sensor and face unlock.

Pulse buyers can opt for 4GB or 6GB of Ram and get 64GB of storage. There's also a microSD card slot for adding another 256GB. The phone goes for €140/£100 (about $125 - $150).

The Pulse Plus offers pretty much identical specs as the base model with the biggest difference being the rear camera, which is a 50MP snapper. It can also be equipped with up to 8GB or RAM and has 128GB of built-in storage. It's priced at €160/£130 ($162 - $171).

The Pulse Pro

The Pulse Pro's main difference is the upgraded 50MP selfie cam and 20W charging, which probably explains why it's only 20 euros ($21) more expensive than the Plus model.

The HMD Vibe (above), the only model coming to the US, will be similar to the Pulse phones, featuring a 6.56-inch 90Hz display. The battery drops to 4,000MhA, it has up to 6GB of memory, and the "AI Dual Camera" system includes a 13MP main snapper. The phone uses a Snapdragon 680 SoC rather than Unisoc's chip and will be priced at $149 when launched in May. It's believed that Verizon will support the handset.

Adobe Photoshop and Firefly receive new AI enhancements, hilarity ensues

Recap: Adobe introduced AI tools to its image editing suite a year ago and now aims to enhance them with better performance and new features. However, the upgrades aren't immune to some of the pitfalls associated with AI, occasionally yielding humorous results.

Adobe's new AI-based foundation model, Firefly Image 3, is now in beta. Available in Photoshop and the Firefly web app, the update enhances the performance of the company's generative AI tools and introduces new methods for creating and editing images.

Introduced last year, Firefly utilizes AI to add or remove objects from pictures, alter surfaces, expand images to wider aspect ratios, and make other modifications. The primary focal points for Firefly Image 3 are style and consistency. Users can now generate pictures in a wider variety of styles or instruct the software to produce new images in a style similar to a reference.

The lighting in realistic pictures should now be more accurate. Furthermore, the new version of Firefly can comprehend longer and more complex prompts more accurately than its predecessor.

Photoshop's generative fill tool, which many users amusingly used to expand iconic album covers last year, gains several new functions from Firefly Image 3. Adobe's most popular image-editing software now features text-to-image generation for the first time, and users can also generate backgrounds for existing pictures.

Moreover, they can enhance details for a sharper image and iterate on creations by requesting similar outputs. Adobe's demonstration video (above) makes Photoshop's new tools appear seamless, but one user recently discovered that they can still lead to unexpected results.

you guys are right AI is taking over the world and i'm terrified ð pic.twitter.com/g4ih8GpZlA

– Nina Nikolic �'� Kerri (@kidkerrigan) April 23, 2024

Developer and animator Nina Nikolic attempted to use generative fill to expand the aspect ratio of a photo of herself in which her shoulder was cut off. However, when she asked the AI model to improve how it filled in the shoulder, it instead generated an entire extra person – a naked man – next to her picture. Another attempt resulted in a hideously deformed giant face.

Also: Low-res video enhancements

In related news, Adobe researchers recently unveiled their work on an AI-based video super-resolution tool. The technology, called VideoGigaGAN, shares the same goal as Nvidia's RTX Video Super Resolution – making low-resolution footage appear cleaner on high-resolution displays. However, Adobe claims its solution adds clearer details for a less blurry result.

Multiple comparisons are available on GitHub.

QDEL technology set to become the premium display replacement to OLEDs by 2026

The bleeding edge: Currently, the highest quality pictures come from displays with OLED screens. The blackest blacks and the bright, vibrant colors it produces are unmatched. The technology comes at a premium, but a stroll through the local department store shows that prices are beginning to come down. As prices start dipping for the latest electronics, we begin wondering about the next generation.

Micro-LEDs are in line to replace OLEDs but need another half decade or so of development before commercialization. Noctiluca notes that micro-LED tech will be ideal for transparent screens and displays viewed outdoors in sunlight, like cell phones and digital billboards. However, there is another display tech called QDEL that is much closer.

Ars Technica opines that QDEL, or quantum dot electroluminescent, might be the next bleeding-edge tech for high-end televisions, monitors, and gadgets with screens. Also known as NanoLED, QDEL is an emissive technology that does not require a backlight – electroluminescent quantum dots emit light directly. These Q-dots are the same as those in high-end QD-OLEDs but are brighter, cheaper, and more resistant to burn-in.

Furthermore, QDEL technology should be available within the next few years, whereas micro-LED won't be ready for consumers until around 2030. Quantum dot supplier Nanosys has been working on the tech for several years. Its potential brand name is NanoLED, but it has gone by other names during development, including QD-EL, QD-LED, EL-QD, and QLED, although Samsung trademarked the latter in 2016, the year before launching its QLED branded TVs.

Nanosys has tentatively targeted 2026 as its window for taking the technology commercial, but as a supplier, it is up to manufacturers like Sharp, Sony, and LG to take advantage of it. So we may not see comsumer sets until later.

"When it comes to consumer gadgets, [I expect] QDEL will most impact TVs, PC displays, and the automotive industry," Omdia Senior Research Director David Hsieh told Ars Tecnica. "If commercialized and mass-produced, QDEL can have a cost-to-performance ratio better than that of OLED, but it would still struggle to compete with LCD-LED on a cost basis."

As an OEM supplier, Nanosys doesn't produce displays, so prototypes that utilize QDEL are scarce, making the company's 2026 timeframe somewhat ambitious. However, it has released some beta-level QDEL tech to allow a handful of companies to experiment.

Digital Trends went to CES 2024 in January, specifically looking for QDEL prototypes. It took a while, and DT correspondent Caleb Denison was about to leave Las Vegas before getting a call from Sharp Display reps. They had heard he was looking for QDEL displays and invited him to view two "secret" prototypes in their CES "backroom" (video above).

Viewing any display in a video is not ideal for getting a good idea of the quality, and QDEL is no exception. However, Denison was impressed and excited by Sharp's small 12.3-inch prototype (shown) and its top-secret 30-inch QDEL display (hidden). He also said that Sharp may have some advancements to show off later this year.

Image credit: PlasmaChem