Safari web apps give frequently used websites many of the qualities normally associated with installed software. A site can receive its own icon, open outside the regular Safari tab interface and appear alongside native apps on supported Apple devices.

The feature is useful for services that do not offer an App Store download, websites used throughout the workday and online tools that feel crowded when surrounded by unrelated browser tabs.

A Safari web app remains a website rather than becoming native Apple software. Its capabilities, interface and offline behavior depend on how the site was developed. Safari supplies the app-like container, while the website continues providing the actual service.

The experience also differs across Apple devices. iPhone and iPad place web apps on the Home Screen, while Mac saves them as separate applications that can open from the Dock, Spotlight or Applications folder.

Safari Web Apps Are More Than Bookmarks

A traditional bookmark saves an address inside Safari. Selecting it opens the page in a normal browser tab with the rest of the browsing session.

A Safari web app creates a more focused entry point. Its icon appears with installed apps, and opening it produces a separate app window or full-screen interface without the usual collection of Safari tabs.

This distinction is particularly valuable for communication tools, project dashboards, streaming services, publications and business systems. A user can switch to the website through the App Switcher, Dock or Spotlight instead of locating the correct Safari window and tab.

Web apps can also support notifications and icon badges when the website provides those features. On iPhone and iPad, supported notifications can appear on the Lock Screen, in Notification Center and on a paired Apple Watch.

The site still controls the quality of the experience. A website designed as a progressive web app may include responsive navigation, offline resources, custom icons and carefully defined window behavior. A basic website may simply open inside a cleaner Safari frame.

How to Create Safari Web Apps on iPhone

Current iPhone software includes an Open as Web App option when adding a website to the Home Screen.

To create one:

1. Open the website in Safari
2. Tap the More button and choose Share. Depending on the selected Safari layout, the Share button may already be visible beside the address field
3. Scroll down and select Add to Home Screen
4. Edit the displayed name if necessary
5. Turn on Open as Web App
6. Tap Add

The new icon appears on the iPhone Home Screen. Selecting it opens the site separately from the main Safari interface, and the web app can be closed through the App Switcher like other iPhone apps.

If Add to Home Screen is missing from the Share menu, scroll to the bottom, select Edit Actions and add it to the available actions.

The chosen name should be short enough to remain recognizable below the icon. Safari may use an icon supplied by the website. When a suitable image is unavailable, the generated icon may use a simplified site image or monogram.

How to Create Safari Web Apps on iPad

The iPad process is similar, although the Share menu reflects the larger Safari interface.

Open the desired website, select Share, choose More if required and tap Add to Home Screen. Turn on Open as Web App before selecting Add.

The resulting icon appears on the iPad where it was created. It can be organized into Home Screen folders and opened through iPad search like other applications.

An iPad web app can be especially useful for browser-based writing tools, office systems, inventory dashboards and educational platforms. Removing the tab bar gives the website more room and makes it easier to use beside another app.

Whether the web app supports multiple windows, external displays or advanced iPad multitasking depends on its design and the capabilities exposed through Safari. Installing a website does not automatically give it every feature available to a native iPad app.

Safari Web Apps Work Differently on Mac

Apple introduced its current Mac web app system with macOS Sonoma 14. Any webpage can be saved as a web app without requiring the website developer to distribute a Mac application.

To create one, open the page in Safari and choose File > Add to Dock. The same command is available through Safari’s Share button.

Safari displays the proposed name and icon before saving the app. After confirmation, the web app appears in the Dock and Spotlight Applications. Its application file is placed inside the Applications folder located in the user’s Home folder.

A Mac web app receives its own window and menu-bar identity. It uses a simplified toolbar containing navigation and sharing controls rather than Safari’s complete tab and bookmark interface.

The web app can remain in the Dock, run automatically as a login item and participate in normal macOS window management. It can also be opened through Spotlight without launching a visible Safari window first.

Mac Web Apps Keep Browsing Separate

A Mac web app operates independently from ordinary Safari browsing. Apple says it does not share ongoing browsing history, cookies, website data or settings with Safari.

When the web app is first created, Safari can preserve an existing website sign-in in many cases. Activity after that point remains inside the web app’s separate data container.

This separation resembles using a dedicated Safari profile, but the web app is even more focused. Searching, following links and changing accounts inside it do not add those pages to the normal Safari history.

The arrangement can be helpful when the same service is used for work and personal accounts. Separate web apps can create clearer boundaries, although account behavior still depends on the website.

Isolation can also explain an unexpected sign-in request. A person may already be authenticated in Safari but encounter a separate session after creating, resetting or reinstalling the web app.

Mac Settings Allow More Control

Mac web apps include settings that are unavailable to ordinary Home Screen bookmarks.

Open the web app, select its name in the menu bar and choose Settings. The application name, starting address and icon can be changed without recreating the app.

Navigation controls can be shown or hidden. Hiding them creates a cleaner appearance for a dashboard that rarely leaves its main page, while displaying them makes sense for sites with deeper navigation.

The title bar can adopt colors supplied by the website. This helps distinguish several installed services but can be disabled when a consistent Mac appearance is preferred.

The Privacy section controls access to the camera, microphone, screen and location. Website data can also be cleared from this area without deleting normal Safari data.

Compatible Safari extensions can be enabled separately for the web app. This prevents every active Safari extension from automatically affecting a focused work tool.

Notifications Make Websites Feel Like Apps

A supported website can request notification permission after it becomes a web app.

On iPhone and iPad, web push notifications use Apple’s notification system. They can appear on the Lock Screen, inside Notification Center and on a paired Apple Watch. They also respond to Focus settings.

Websites must request permission after a direct user action, such as selecting a notification or subscription button. Simply adding a site to the Home Screen does not automatically authorize alerts.

Supported sites can place a count on the Home Screen icon through web badging. This may represent unread messages, pending tasks or new articles, depending on the service.

Mac web apps can send notifications and display unread badges on their Dock icons. Permission should be granted from inside the web app rather than through the same site in Safari if the alerts are intended to belong to the installed app.

Each web app appears by its chosen name in Apple’s notification settings. Alerts, sounds, badges and preview behavior can then be controlled much like those of downloaded software.

Web Apps Do Not Automatically Work Offline

An installed icon does not mean the website has been downloaded completely.

Most web apps continue to require an internet connection because their pages, account data and services remain on remote servers. If the connection disappears, the app may show an error or only display information already stored temporarily.

Websites designed for offline use can cache interface elements and selected content through supported web technologies. The developer decides what remains available without a connection and how changes synchronize later.

This makes offline behavior inconsistent between sites. One writing service may continue editing disconnected documents, while another cannot move beyond its sign-in page.

Anyone depending on a web app during travel should test it in Airplane Mode before assuming that important information will remain accessible.

Links Can Return to Safari

Links belonging to the same website usually remain inside the web app when they fall within its intended navigation area.

External links may open in Safari because they lead beyond the web app’s domain or configured scope. Authentication systems, payment providers and support pages can also move temporarily into a browser window.

This behavior protects the focused identity of the web app. A project-management tool does not need to become a complete browser simply because a task contains a news link.

A poorly configured website may open more links externally than expected. That behavior is generally controlled by the site rather than by a Safari setting available to the user.

Safari Web Apps Do Not Sync Automatically

A web app icon appears only on the Apple device where it was created.

Adding a website to an iPhone Home Screen does not install the same web app on an iPad or Mac, even when every device uses the same Apple Account. Each installation must be created separately.

Safari bookmarks, Reading List items and open tabs can synchronize through iCloud, but web app installations remain local. Mac settings and custom icons also do not automatically transfer to another Mac.

This local behavior allows each device to maintain a different collection. A detailed analytics dashboard may belong on a Mac and iPad, while a compact travel service may only need an iPhone icon.

Removing a Web App Leaves the Website Available

On iPhone or iPad, touch and hold the web app icon and select the removal option. Deleting the icon does not cancel the website account or remove the service from the internet.

A Mac web app can be deleted by opening the Applications folder inside the user’s Home folder and moving the app to the Trash. Removing it from the Dock alone only removes the shortcut.

Website data may remain until it is cleared through the web app’s privacy settings or removed with the app, depending on the platform and current software behavior.

Saved documents stored exclusively by the online service remain subject to that service’s account and retention policies. Removing the web app should not be treated as deleting information held by the website.

Apple Vision Pro Uses a Different Safari Model

Safari on Apple Vision Pro can open multiple spatial windows, synchronize bookmarks and tabs through iCloud, save pages to Reading List and place separate browser windows around the user’s space.

Apple’s current Vision Pro guide does not provide the same Add to Home Screen or Add to Dock installation process documented for iPhone, iPad and Mac.

A website can still feel application-like inside its own Safari window, particularly when positioned as a persistent part of a workspace. That does not make it an installed Home View app.

Apple Watch also does not provide a general Safari web app installation system. It can, however, display supported notifications sent by an iPhone Home Screen web app.

Choose the Starting Page Carefully

A Safari web app opens from a defined starting address, making the selected page important.

Installing a service from its general homepage may produce an app that repeatedly opens promotional content or a sign-in screen. Creating it while viewing an inbox, dashboard, calendar or project page can provide a faster route to the area used every day.

Mac users can change the Application URL later through the web app settings or use Set to Current Page after navigating to a better starting location.

The most effective Safari web apps begin where the work actually happens. A publication can open directly to its latest articles, a company tool can start on its task board and a cloud service can lead to the folder used most often.

That small choice determines whether the icon functions as another website shortcut or becomes an app-like destination that removes several steps from a repeated Apple device workflow.

The post Safari Web Apps Turn Websites Into Apple Device Apps appeared first on AppleMagazine.

OpenAI family of devices plans are no longer limited to an experimental gadget hidden inside a design studio. OpenAI President Greg Brockman has confirmed that the company’s first consumer device will begin a broader series intended to reconsider how people interact with artificial intelligence.

The confirmation gives OpenAI’s hardware strategy a clearer shape without revealing the first product itself. Brockman said the company is not ready to discuss the device in detail, but described it as the first in a series that will rethink what devices should be when AI can understand language, observe context and complete tasks.

That ambition extends well beyond placing the ChatGPT app on another piece of hardware. OpenAI wants to reduce the menus, buttons, model selectors and separate applications that currently stand between a person and an AI assistant.

For Apple, Google and other established platform owners, the implication is significant. ChatGPT currently reaches consumers through devices and operating systems controlled by other companies. An OpenAI-designed product would give the company direct control over the microphones, cameras, sensors, interface and account experience through which its models interact with people. 56000

OpenAI Family of Devices Begins With io

OpenAI’s move into consumer hardware accelerated in 2025 when it acquired io Products in a transaction valued at approximately $6.5 billion.

The startup had been founded by former Apple design chief Jony Ive alongside Scott Cannon, Evans Hankey and Tang Tan. Its team included hardware engineers, designers, manufacturing specialists and product developers, many of whom had spent years working together at Apple.

OpenAI and Ive had already been collaborating before the acquisition. Their early ideas developed into physical designs, convincing the partners that they needed a dedicated company capable of engineering and manufacturing complete products rather than another software prototype.

The io team officially merged with OpenAI in July 2025. Ive and his design collective LoveFrom remained independent while assuming extensive design and creative responsibilities across OpenAI.

The structure combines OpenAI’s models and software research with a group experienced in turning new interface concepts into mass-produced consumer products. Ive’s previous work helped shape the iMac, iPod, iPhone, iPad, Apple Watch and other devices that established Apple’s modern design identity.

OpenAI has continued recruiting hardware expertise from Apple. Paul Meade, who led hardware development for Apple Vision Pro and worked on the company’s smart-glasses efforts, reportedly joined OpenAI’s hardware organization in 2026.

OpenAI also placed product-development veteran Peter Welinder in a leadership role connected to the io division. The resulting organization is not structured like a small accessory project. It is being assembled to develop products, create manufacturing relationships and support more than one form factor.

The First Device Remains Unannounced

OpenAI has not publicly confirmed the name, appearance, price or specifications of its first device. It has also not presented the product as a smartphone, pair of glasses, wearable pin or conventional smart speaker.

Recent reporting describes a portable, screen-free device designed primarily for use inside the home. The reported product resembles a compact speaker but includes a rechargeable battery, allowing it to move between rooms rather than remain attached to one location.

A camera and additional sensors would help the device understand its surroundings. Mechanical elements could allow parts of it to move, creating a more expressive physical presence than a stationary speaker waiting for a wake word.

The reported design would rely heavily on OpenAI’s latest voice technology. Instead of requiring a person to wait for the assistant to finish speaking, the system could listen and respond more naturally during overlapping conversation. Access to personal information, including messages or email when authorized, could give it enough context to handle requests rather than merely answer questions.

These details remain reported plans rather than announced specifications. OpenAI could change the hardware, features or release schedule before introducing the product.

The company is reportedly targeting a presentation during 2026 and a commercial release in 2027. Fortune separately reported that the initial design has been finalized, although development of the larger product family continues.

Why OpenAI Wants Hardware of Its Own

ChatGPT already operates across iPhone, iPad, Mac, Android and Windows devices. That wide distribution gives OpenAI access to an enormous audience without requiring it to manufacture screens, batteries or processors.

It also leaves crucial parts of the experience under the control of Apple, Google and Microsoft.

An operating system determines when ChatGPT can use a microphone, whether it can run continuously, how notifications appear and which applications it can control. Platform owners can limit background access, change default-assistant rules or promote their own AI systems more prominently.

Those restrictions matter more as ChatGPT develops from a chatbot into an agent. Answering a question inside an app requires relatively little control over the device. Completing a task across messages, calendars, files, cameras and connected appliances requires deeper access.

OpenAI hardware could be designed around those capabilities from the beginning. The assistant would not need to behave like one application among dozens because it could become the primary interface.

Brockman has described the current ChatGPT interface as too complicated, particularly as new models and tools add more controls. OpenAI’s objective is to make much of that interface disappear.

A successful device could allow someone to describe an outcome instead of opening several applications and completing each step manually. The user might ask the assistant to reorganize a day after a delayed flight, contact the relevant people and adjust reservations. The interface would be the request and the resulting confirmation rather than a sequence of taps.

A Family Creates More Context Than One Gadget

Calling the project a family of devices changes the strategy. One experimental product would need to prove its value independently. A coordinated group could distribute different sensors and functions across the places where people already live and work.

A home device could provide shared access in kitchens and living spaces. A personal wearable could preserve private conversations while moving outside. A future mobile product could handle visual information, navigation and communication without reproducing the application grid of a smartphone.

OpenAI is reportedly developing approximately five hardware concepts, with the screen-free home companion receiving priority. The company has also explored wearables and home robotics, although none of those later products has been announced.

The devices would become more useful if they shared the same ChatGPT identity, memory controls and task history. A request started at home could continue through another product without requiring the user to repeat the people involved, the desired result or the decisions already made.

That continuity is one of the strongest advantages Apple currently receives from controlling a complete ecosystem. An iPhone, Mac, Apple Watch, AirPods and HomePod can exchange information because Apple designs the devices, operating systems and cloud services together.

OpenAI appears to be pursuing a different center of gravity. Instead of making the operating system coordinate applications, it could make the AI assistant coordinate specialized devices.

ChatGPT Will Also Expand Through Other Companies’ Hardware

OpenAI’s hardware work does not mean the company intends to reserve ChatGPT for its own products.

Brockman has said OpenAI is interested in deeper integrations with smart glasses, wearables and other devices through its application programming interfaces. The company wants ChatGPT to become an interface available across hardware categories, including products manufactured by third parties.

That creates a two-track strategy. OpenAI can use its own devices to demonstrate the experience it considers ideal while allowing other manufacturers to connect their cameras, microphones and sensors to ChatGPT.

Third-party distribution would help OpenAI reach consumers who do not want another dedicated device. It would also reduce the risk of depending entirely on the success of one hardware launch.

The difficult part will be keeping those experiences consistent. A wearable may have limited battery capacity, a camera may require visible privacy controls, and a speaker may depend on cloud processing when a connection is available. OpenAI will need to define what its assistant can do reliably across products with very different technical limits.

Early AI Hardware Changed the Expectations

The first generation of dedicated AI gadgets established a warning for every company entering this market.

Humane’s AI Pin and the Rabbit R1 attracted attention by promising an alternative to application-based computing. Their launches also exposed problems involving response delays, battery life, incomplete features and tasks that remained easier on a smartphone.

Those products demonstrated that access to a language model does not automatically justify a new object. Consumers need to understand why they should carry, charge and maintain another device when ChatGPT, Gemini and other assistants already operate on phones.

OpenAI has several advantages those earlier companies lacked. It controls one of the most widely used AI services, operates the models behind the interface and has the resources to improve the software after launch. The io acquisition gives it experienced design and manufacturing leadership rather than requiring it to assemble a hardware organization gradually.

The standard will consequently be higher. A well-designed enclosure and natural voice will not be enough if the device cannot complete dependable tasks more efficiently than an iPhone or Android phone.

Apple’s Lawsuit Could Affect the Schedule

OpenAI’s expansion beyond Apple devices has also produced a direct legal confrontation with Apple.

Apple filed a federal lawsuit in July 2026 accusing OpenAI, io and former Apple employees of misappropriating confidential hardware information. The complaint alleges that trade secrets connected to unreleased products, engineering processes and suppliers were taken for the benefit of OpenAI’s hardware business.

OpenAI has denied having any interest in Apple’s trade secrets and said it was unaware of evidence giving the complaint merit. The allegations have not been decided by a court.

Apple is seeking restrictions on the use of its confidential information, along with the return of materials and financial damages. The case could require OpenAI to separate employees from parts of the project, replace suppliers or repeat engineering work if disputed information is found inside the development process.

It does not automatically establish that OpenAI’s reported product copies an Apple device. The first OpenAI design is described as substantially different from Apple’s current hardware, but the legal dispute could still affect when it can be manufactured or sold.

The First Product Will Define the Rules for the Rest

The most important details at OpenAI’s eventual hardware presentation may have little to do with the object’s shape.

A camera-equipped device that learns from activity inside a home will need unmistakable recording indicators, understandable data-retention controls and a simple method for temporarily disabling its sensors. Owners will need to know which information remains on the device, what reaches OpenAI’s servers and which memories can be reviewed or deleted.

OpenAI will also have to explain whether the product requires a paid ChatGPT subscription, how much it can do without an internet connection and whether multiple members of a household receive separate identities. Those choices will determine whether it behaves like a personal assistant, a shared appliance or a new subscription terminal.

Support for messages, calendars, music and connected appliances will require cooperation from outside companies. A device cannot replace frequent phone interactions if it remains unable to enter the services where those interactions occur.

The first model will therefore establish more than OpenAI’s industrial-design language. It will set the privacy, account, subscription and developer rules that every later member of the OpenAI family of devices inherits.

The post OpenAI Family of Devices Takes ChatGPT Beyond Apple appeared first on AppleMagazine.

Global smartphone revenue rose 7% year-on-year in Q2 to an historic $109 billion, the highest second-quarter total on record even as shipments declined, Counterpoint Research revealed.

The disconnect between rising revenue and falling volumes was driven by average selling prices (ASPs), which rose 17% to a record high of $400, fuelled by sustained demand for premium devices and memory-driven price hikes across Android portfolios.

Apple was the standout performer, with revenue up 22% to the brand’s highest-ever Q2 total.

The growth was supported by a 13% rise in shipments alongside an 8% increase in ASP, pushing Apple’s revenue share to a record 49%.

Counterpoint Research director Tarun Pathak said Apple’s performance was underpinned by sustained demand for the iPhone 17 line-up, particularly the base 17 and 17 Pro Max, which kept the sales mix skewed toward premium smartphones.

Unlike rivals which pushed through steep price increases, Apple largely held pricing steady, a move Pathak said reflects the company’s ability to absorb rising bill-of-materials costs and stay comparatively insulated from the industry’s memory chip crunch.

He cautioned Apple would likely raise prices in coming quarters, but believes the pricing discipline so far has strengthened its competitive position, enabling growth in both value and volume.

China, Europe and emerging markets were strongholds, as Apple’s stable pricing improved its relative value proposition against widespread Android price hikes.

Samsung ranked second with a 16% revenue share, with both revenue and shipments up 9% helped by steady Galaxy A-series demand and continued momentum from the Galaxy S26 series, along with strong growth in North America and the Middle East and Africa.

Xiaomi posted the steepest shipment decline among the top five brands, down 26%, dragging revenue down 17% despite a 13% rise in ASP as the company pushed toward higher-value devices.

Oppo and vivo also saw revenue fall, down 10% and 11%, respectively.

Senior analyst Shilpi Jain said the market has entered a phase where value, not shipment volume, is the primary growth driver, with OEMs increasingly passing higher component costs on to consumers.

With memory shortages showing no near-term signs of easing, the research company expects further price increases and steeper shipment declines across the industry in the second half of 2026.

The post Smartphone revenue hits $109B as Apple claims record share appeared first on Mobile World Live.

Macworld

At a glance

Expert's Rating

Pros

• Fast 80Gbps performance
• Four NVMe slots
• RAID 0, 1, 4, 5, and 10
• Built solid as a tank

Cons

• RAID is software only
• Only 20Gbps performance for individual drives
• Minor fan noise

Our Verdict

Run in RAID 0, the 4M2 Ultra is capable of over 7GBps reads. Good stuff for this handsome, 4-bay 80Gbps NVMe enclosure..

Price When Reviewed

This value will show the geolocated pricing text for product undefined

Best Pricing Today

Best Prices Today: OWC 4M2 Ultra 80Gbps 4-disk RAID enclosure

Retailer

Price

Check

Price comparison from over 24,000 stores worldwide

Product

Price

Price comparison from Backmarket

OWC’s 80Gbps 4M2 Ultra is a four-slot RAID enclosure that lets you build your own high-capacity, super-fast external storage using NVMe SSDs.

Depending on the RAID mode you use, you can saturate the 80Gbps Thunderbolt 5/USB4 bus to the tune of 7GBps sequential transfers. You can also safeguard your data with mirroring or parity, which a single external SSD enclosure won’t allow.

What are the OWC 4M2 Ultra’s features?

The 4M2 Ultra is a black, mostly aluminum box measuring around 4.75-inches tall, 4.5.inches deep, by 2.35-inches wide. Unlike the 40Gbps 4M2, the sides are finned to create more heat-shedding surface area. Nice. On the back are two Thunderbolt 5 ports (one marked as the computer port), a Kensington lock port, and the AC adapter jack…

You didn’t think you could run four SSDs on bus power alone, did you? Well, being curious and considering that Thunderbolt 5 is required to deliver 140 watts (240 watts optionally) I pulled the power jack with the 4M2 Ultra attached to my M4 Max Studio. No blue power light. Sigh.

I was a bit surprised to see the 4M2 Ultra open like a flower when I removed the screw-down backplate. The backplate holds the two side plates (keyed for reattachment) in place, so both fell to the side, revealing two 2280/2242 M.2 slots under each.

Thermal strips are pre-attached to the side plates, which is thoughtful, though mine didn’t have any protection on them. Before I realized that, I’d minimally damaged one by stacking one side plate on the other. Easily repairable, so no big deal.

The 4M2 Ultra’s dual PC board, chips and fan are sandwiched between the two dual-drive bays. The front of the 4M2 Ultra is perforated to allow air flow through the unit to the vents on back.

You can hear the fan when it kicks in. It’s not loud, but if such things bug you, you’ve been informed. I find the sound reassuring in that the 4M2 Ultra isn’t allowing your SSDs to fry.

The Intel JHL9480 (Barlow Ridge) controller allocates each of the 4M2 Ultra’s M.2 slots a single PCIe 4.0 lane, which amounts to around 2GBps of bandwidth per slot. Hence, while you could run the four SSDs singly, you’ll only get 80Gbps (or 40Gbps) performance with all four drives combined in RAID 0 (fastest) or RAID 10 (next fastest).

To operate the Express 4M2 Ultra in RAID, you need software. OWC provides a license for its handy cross-platform SoftRAID, though the inlcuded Pro license expires three years from the date of purchase. However, the core functions (operation and rebuilding arrays) remain eternally functional.  

If expiring software makes you nervous, there’s always macOS RAID or Windows Storage Spaces. Sadly, Windows Storage Spaces is pathetically slow, but macOS RAID 0 was competitive with SoftRAID

The 4M2 Ultra is warrantied for two years, however, given the quality of the construction it should last well beyond that.

How fast is the 4M2 Ultra?

When running four SSDs in RAID 0, the 80Gbps 4M2 Ultra is a very fast reader, up to 7GBps (Windows). It’s a considerably slower writing at a little over 2GBps according to most of our tests.

The 4M2 Ultra is also significantly faster in every RAID mode than the older 40Gbps 4M2 — hardly surprising given twice the older unit’s 40Gbps bandwidth. Note that all RAID tests were done while using OWC’s SoftRAID in RAID 0.

AmorphousDiskMark had the 4M2 Ultra at 6.35GBps reading, and 2.35GBps writing.

Disk Speed Test from Blackmagicdesign wasn’t quite as giddy as AmorphousDiskMark, and for some reason, likely a much smaller data set, showed 5GBps writing and only 2GBps reading.

I also ran the 4M2 Ultra through PCWorld’s battery of tests for comparison purposes. However, the other drives shown: the Satechi DotDisk, TerraMaster D1 SSD Pro, and OWC’s own 1M2 are 80Gbps single drive enclosures, which makes these comparisons a tad unfair. But the charts do show what you stand to gain or lose by going the RAID route.

The 4M2 Ultra did well in some CrystalDiskMark 8 (the code base for AmorphousDiskMark) tests, while in others, not so much. And with four SSDs to stripe writes across, the 4M2 Ultra’s write performance is a tad disappointing.

The 4M2 Ultra lagged a bit in CrystalDiskMark 8’s 4K tests, but it’s still fast in the grand scheme of external storage.

In PCWorld’s real life 48GB transfers, the 4M2 Ultra was quite close to the competition, and even tied for first in one test.

PCWorld’s 450GB write test proved the 4M2 Ultra to be a very fast solution when transferring very large files using Windows Explorer, but didn’t perk up, as is usually the case, with the fantastic FastCopy utility.

All told, the 4M2 Ultra is fast, and a great improvement over the older 40Gbps 4M2. However, it’s not as fast as single enclosures, so its most salient strength is combining smaller SSDs into a single large volume, up to 32TB with 8TB SSDs.

Note that while testing, it was discovered that Phison E26 controllers apparently require more power than my M4 Max Studio can provide under heavy loads at 80Gbps (10Gbps and 40Gbps are fine). Using these in the 4M2 Ultra was a rocky experience–they would disconnect under load. This did not occur with the 4M2 Ultra on PCWorld’s Wintel test beds. If you’re a Wintel user who has E26 SSDs, you’re safe. Have an 80Gbps Mac? Avoid them.

Should I buy the 4M2 Ultra?

If you want fast, super vast external storage, the 80Gbps 4M2 is a great way to create speedy large volumes from smaller SSDs. Just avoid E26-based SSDs.

Tim Cook’s last earnings call as the CEO of Apple covered record revenue across iPhone, Mac and services despite ongoing supply constraints for memory chips.

Revenue in its fiscal Q3 2026 (the period from 29 March to 27 June) was $109.4 billion, up 16.4% year-on-year.

Cook (pictured) took the helm from Apple co-founder and chief Steve Jobs in 2011, when the company posted annual revenue of $108 billion.

“It’s a special privilege to be part of people’s lives in lasting and meaningful ways, helping them to create, connect and experience the world around them,” Cook said on the earnings call.

“The transition is going seamlessly, and I am beyond excited for John [Ternus] to step into his new role and lead Apple into its next era.”

Tariffs
Apple’s booked a $0.11 per-share boost from tariff refunds, which added roughly two percentage points to gross margin after the US Supreme Court struck down a tranche of the Trump administration’s fees earlier this year.

“Last year, we made a $600 billion commitment to the US over four years and now, as we said before, we plan to reinvest the tariff refunds we’ve received into the US,” Cook said.

“We’re pleased with the progress we’ve already made advancing the American supply chain.”

Cook highlighted Apple sourcing more than 100 million components from a TSMC production facility in the US state of Arizona as an example.

“We could not be more pleased with how that fab has ramped and is producing for us.”

Siri AI
He noted positive developer reviews of Siri AI, which was introduced at WWDC in June.

“I think it’s a very big idea to have AI that’s private, that’s based on your personal context and that’s integrated across the operating system.”

Siri AI is scheduled to be released in September alongside iOS 27 and the iPhone 18 line-up.

Cook noted Apple is working closely with the European Commission to make the assistant available on devices in the European Union and said the company recently received approval to ship some of the original features of Apple Intelligence for use in China.

“We’re working now through the rollout of those and there will be more work required down the road for Siri AI,” he said. “But we’re at the front end of that.”

Supply chain
Apple flagged persistent supply constraints heading into the current fiscal quarter, forecasting revenue growth of 9% to 11%.

Cook described the memory chip shortage driving the pressure as a “hundred-year flood,” warning Apple expects to keep paying more after three consecutive quarters of price increases, albeit with some inventory offset and reductions in other components.

Numbers
Net income rose 27.1 per cent to $29.8 billion.

iPhone revenue rose 21.7% to $54.3 billion, which Apple branded a record for the period.

Greater China revenue rose 22.4% to $18.8 billion.

Service revenue was $30.7 billion and Mac revenue rose 28.7% to $10.3 billion.

The post Apple CEO swan song ends on high note appeared first on Mobile World Live.

Samsung Electronics booked an operating loss at its mobile division in Q2 due to elevated component costs and revealed plans for efficiency initiatives at the unit, as it registered another all-time high in overall quarterly earnings driven by memory sales.

Its Mobile eXperience (MX) division and associated Networks Business recorded an operating loss of KRW700 billion ($480 million) compared with a profit of KRW2.8 trillion in Q2 2025. Revenue was up 14% to KRW33.2 trillion

Samsung noted despite “solid sales” of its flagship Galaxy S26 devices and “strong momentum” for its lower-end Galaxy A series, “earnings declined due to elevated component cost pressures across the industry”.

The company plans to continue its focus on providing AI experiences on its devices and “drive overall market share growth based on flagship-first expansion centred on the new Galaxy Z8 and S26 series”.

It also flagged plans to launch smart eyewear and “pursue efficiency initiatives to mitigate the impact of rising costs”.

For networks, it pointed to improved earnings supported by increased sales outside of its home market of South Korea.  

While the company suffered from rising component costs for its devices, Samsung’s memory chip business recorded what it described as an all-time high in quarterly revenue and operating profit.

The Device Solutions division, which houses the memory unit, booked an 89% increase in operating profit to KRW89.2 trillion on revenue up 357% to KRW127.5 trillion.

Across the whole of Samsung Electronics, revenue was up 130% to KRW171.5 trillion and net profit grew from KRW5.1 trillion to KRW71.6 trillion.

The post Samsung experiences both sides of memory boom appeared first on Mobile World Live.

Qualcomm reported fiscal Q3 results which beat Wall Street’s revenue expectations but missed on profit, as the chipmaker continued to grapple with the impact of a memory supply crunch weighing on the smartphone sector.

The company posted revenue of $9.9 billion for the quarter ended 28 June, down 4 per cent from $10.4 billion a year earlier but ahead of the roughly $9.7 billion analysts had expected. Net income during the period was $2 billion, down 25% year-on-year.

Handset chip sales fell 20% to $5.1 billion, which Qualcomm CEO Cristiano Amon (pictured) stated reflected a bottoming out in the China market after OEMs pulled back on orders amid memory supply uncertainty.

“We actually don’t expect fundamental changes in the premium tier and the higher tier volume,” Amon said on the company’s investor call.

Going forward, the executive expects Apple-related revenue to decline sharply and faster than previously anticipated, after Qualcomm’s share of components in the upcoming iPhone range came in well below an earlier estimate of around 20%.

Amon said the company is offsetting the loss with growth elsewhere, including its emerging data centre business, which it expects to generate $5 billion in revenue by fiscal 2027.

“We expect Android to grow and significantly offset the reduction in Apple, and then data centre revenue that starts to ramp in the fourth quarter,” CFO Akash Palkhiwala said.

Amon said its Q3 results reflected “solid execution of our growth strategy” despite the challenging environment.

He noted Qualcomm is taking steps to protect margins going forward, including raising prices across its chip portfolio from 1 September, as component costs continue to climb.

Automotive remained a bright spot, with revenue reaching $1.59 billion which marked a 61% increase and the 23rd consecutive quarter of double-digit year-on-year growth for the segment. Qualcomm reiterated a target of $10 billion in annual automotive revenue by fiscal 2029.

IoT revenue was $1.83 billion, representing a 9% increase. Combined automotive and IoT revenue climbed 28%, underscoring the company’s push to diversify beyond its traditional handset business as it also builds out a data centre and custom silicon strategy with a leading hyperscaler.

The post Memory crunch bites Qualcomm handset chip sales appeared first on Mobile World Live.

Google expanded its Play Age Signals API to all developers globally, part of a broader push to help parents and app developers deliver safer, age-appropriate experiences on Google Play.

Currently available in Brazil, the tool will be introduced to users in Australia and Canada by mid-August, with a full global rollout planned later this year, according to Paul Feng, Google Play’s vice president of product management.

The Play Age Signals API is designed as a privacy-preserving tool that lets parents share their child’s age range, such as 16-17, directly with apps, while also allowing adults to share their own age when prompted by developers.

Developers then receive these signals to tailor in-app safety settings and content to match their audience, rather than applying uniform rules across all apps regardless of content type.

Feng stated the approach gives developers flexibility to choose protections appropriate to their app’s nature, noting a weather app shouldn’t require the same safety settings as an entertainment or media app.

For parents, age-sharing controls are centralised within the Google Family Link app rather than requiring configuration across individual apps.

Once a parent chooses to share a child’s age range, all Play apps using the API can receive that signal, letting children access age-appropriate content without manual setup in each app. Google said age ranges are never shared by default, and parents can update or disable the settings at any time.

The API builds on Google Play’s existing safety infrastructure, including mandatory safety standards for apps designed for families, ongoing app reviews and scanning, and tools such as Restrict Minor Access within the Play Console, which helps developers manage who can discover their apps.

Family Link continues to serve as a central dashboard for parents to manage screen-time limits, PIN-based content filters and app download approvals.

Google positioned the global expansion as an addition to its existing safety suite, aimed at helping parents and developers collaborate to make Google Play safer and more trustworthy for families.

The expansion comes as lawmakers and regulators around the world are increasingly pressuring app stores to provide better protections for minors.

Earlier this year, Apple pushed out age verification rules across a range of its services in the UK, responding to pressure from the government around ensuring safety on the internet for younger users.

The post Google expands Play Age Signals API globally appeared first on Mobile World Live.

By controlling a type of polarized light widely used in quantum optics, researchers have transformed subtle optical patterns in our peripheral vision into vibrant, easily perceptible shapes. Led by Dusan Sarenac at the University of Waterloo, Canada, the team showed for the first time that the “Boehm’s brushes” perceived by participants can be altered by adjusting the topology of spin-orbit coupled light. Their results, reported in Proceedings of the National Academy of Sciences, could lead to promising new techniques for checking our eye health.

For the most part, human eyes are incapable of detecting polarization in visible light. But in some rare cases, structures inside the eye itself can scatter light in different ways depending on its polarization, creating subtle yet perceivable “entoptic” patterns. Perhaps the best-known examples of these are “Haidinger’s brushes”: hourglass-shaped patterns that can form in our central vision, but only when viewing linearly polarized light at blue wavelengths.

In 1940, Swiss biophysicist Gundo Boehm discovered another example: a two-lobed, bowtie-shaped entoptic pattern that forms when viewing linearly polarized light in the peripheral vision as it scatters from subcellular structures in the inner retina. Unlike Haidinger’s brushes, this lesser-known effect is driven by angular variations in scattering strength relative to the polarization direction.

But as Sarenac explains, these ethereal shapes are more than just a curiosity. “Because retinal disease may alter a person’s ability to perceive the pattern, researchers have long considered whether Boehm’s brushes could serve as a biomarker of retinal health,” he says. “However, their weak visibility has limited practical use.”

Whereas previous studies simply asked participants to passively observe the effect, Sarenac’s team considered what would happen if the light’s polarization was constantly varied. To explore this possibility, the researchers created a beam of spin-orbit coupled light, in which polarization is linked to the light’s orbital angular momentum – a property describing how the wavefront twists as the beam travels – so that the polarization direction rotates steadily across the beam.

In quantum optics, these structured waveforms are already being widely explored for their possibilities in imaging, communication and information processing. But for the first time, Sarenac’s team considered how our eyes could actually perceive this light. “We engineered the spatial polarization of the light to match the symmetry of the retinal scattering response,” Sarenac describes. “This allowed many weak local contributions to reinforce one another.”

The researchers then projected this structured light through a ring-shaped aperture, targeting specific regions in the periphery of participants’ retinas, which could also be adjusted to target varying distances from the centre of vision. Using an automated test, they measured how much contrast participants needed to reliably distinguish a Boehm’s brush from the background at these varying distances.

Based on the subject’s responses, Sarenac’s team determined that the use of spin-orbit coupled light not only made Boehm’s brushes appear far more vivid to the participants, it also transformed the bowtie shape into a multi-lobed pattern, with the number of lobes varying depending on the topology of the light’s twisting polarization.

“Across 11 participants, detection thresholds improved exponentially away from the centre of vision, and the pattern became robustly visible at approximately one degree of retinal angular distance,” Sarenac describes. These results reveal a fascinating link between human vision and quantum mechanics, which the team hope could be harnessed in practical healthcare. With further studies involving patients with retinal disease, they now envisage eye tests where retinal conditions could be diagnosed in people who can’t clearly perceive vibrant, multi-lobed Boehm’s brushes in their peripheral vision.

“In the longer term, these patterns may provide a non-invasive functional probe of retinal integrity because their shape and visibility can be measured across different retinal locations,” Sarenac says.

The post Structured light could deliver a new method for diagnosing eye disease appeared first on Physics World.

Cardiac arrhythmia – a disorder in which the heart beats too fast, too slowly or irregularly – affects millions of people worldwide. Currently, arrhythmias are treated using pacemakers to regulate the heart’s rhythm, but traditional pacemakers require invasive surgical implantation and intracardiac leads that risk infection and tissue injury. As an alternative, a research collaboration headed up at Massachusetts Institute of Technology (MIT) and the University of Southern California (USC) has developed a non-invasive pacemaker that’s worn like a sticker on the chest and stimulates the heart using ultrasound.

“Ultrasound offers a unique combination of deep tissue penetration, spatial focusing and non-invasive delivery,” explains first author Chen Gong from USC. Previous approaches for ultrasound-based cardiac pacing, however, lacked reliability or required the use of microbubbles. In this new study, Gong and collaborators employed a technique called sonogenetics, in which cardiac cells are genetically altered to increase their sensitivity to ultrasound. Once exposed to ultrasound waves, the engineered cells trigger the opening of ion channels that let in calcium, which signals the cells to squeeze and beat.

“Our central motivation was to address the limitations of conventional pacemakers, while preserving precise control of cardiac rhythm,” says Gong.

The team first examined whether ultrasound could modulate cardiac cells in vitro, using sonogenetically engineered human heart muscle cells. Ultrasound stimulation caused almost three-quarters of the engineered cells to beat in synchrony with the ultrasound waves. In contrast, unaltered cells did not exhibit such behaviour. Importantly, ultrasound exposure did not impact the cells’ expression of biomarkers for inflammation, myocardial injury or oxidative stress, demonstrating the safety of this approach.

The non-invasive ultrasound pacemaker

The researchers designed a sonogenetics-based non-invasive ultrasound pacemaker (NUP) capable of real-time imaging and steerable stimulation, describing the device in Nature Biomedical Engineering. The postage stamp-sized device, which sticks to the skin via a bioadhesive interface, incorporates a 64-channel phased-array transducer, as well as data acquisition, wireless transmission and power modules.

To assess whether this prototype NUP could achieve non-invasive cardiac pacing in sonogenetically engineered rats, they attached the device onto the rats’ chests and applied ultrasound stimulation through the chest wall to the heart.

At an acoustic pressure of 2 MPa and a 40 ms pulse duration, the NUP increased the rats’ heart rates from 240 to 360–540 bpm, with a higher pulse repetition frequency leading to increased heart rate. Upon stopping the ultrasound stimulation, the heart returned to its natural rhythm. In control rats without the genetic modification, NUP could not control the heart rate effectively.

The team demonstrated that the NUP could electrically steer the ultrasound beam focus with a spatial precision of less than 1 mm, enabling targeting of different regions of the heart to induce chamber-specific pacing. NUP also successfully treated cardiac arrhythmias induced in the engineered rats, returning their heart rates to normal levels. In contrast, ultrasound stimulation of control rats did not effectively treat their arrythmias.

“A major objective of this work was to move beyond proof-of-concept stimulation and demonstrate a wearable system that could realistically support daily use,” explains Qifa Zhou, group leader at USC.  “The current prototype integrates imaging, stimulation, wireless communication and battery-powered operation into a wearable format.”

Clinical potential

For future clinical use, the NUP device needs to automatically detect the wearer’s heart rate, locate heart chambers and deliver stimuli. To achieve this, the researchers used a cloud-based, artificial intelligence (AI)-powered imaging feedback loop that determines the heart rate and stimulation coordinates, and steers the ultrasound beam to target sites.

To test its feasibility for human-scale applications, they assessed the acoustic energy penetration of the device. Using simulations and experiments on a pig heart beneath layers of tissue, they confirmed that, even after tissue attenuation, NUP could deliver sufficient ultrasound pressure (approximately 2 MPa) to a human heart to enable stimulation at clinically relevant pacemaker sites.

The team also assessed the biosafety of the NUP approach, finding that the delivered ultrasound energy remained below approved safety limits and generated minimal thermal effects during and after stimulation. They also confirmed that the sonogenetic engineering induced no off-target effects, immune responses or pathological changes in the rats.

“In the long term, we are optimistic about applying sonogenetics in humans,” explains co-author Gengxi Lu from MIT. “Sonogenetics is different from gene editing – it does not aim to rewrite a patient’s DNA sequence, but instead enables target cells to temporarily or controllably express ultrasound-responsive proteins. At the same time, extensive clinical validation will still be needed to establish safety and long-term effectiveness.”

The team is now focusing on clinical translation and validation of the NUP technology, which includes improving gene delivery strategies, validating the system in larger animal models, and developing closed-loop control approaches that combine real-time physiological sensing with adaptive stimulation. There’s also much engineering work to be done, such as device miniaturization, and improving battery life, skin coupling and motion robustness.

“For cardiac pacing, we envisage that the final goal of NUP technology is to be a permanent alternative to a long-term implanted pacemaker,” Xuanhe Zhao from MIT tells Physics World. “More broadly, we are interested in expanding ultrasound-enabled bioelectronic medicine beyond cardiac pacing toward other organs and therapeutic applications where non-invasive, spatially precise modulation could have clinical impact.”

The post Wearable pacemaker uses ultrasound to control heart rhythm appeared first on Physics World.

A new contact-free way of rotating delicate samples, such as biological cells, undergoing three-dimensional (3D) optical microscopy increases the image resolution and could reveal more details about microscale structures and processes.

Current optical microscopes provide extremely good image resolution in a single plane, but since depth information is often lacking, samples are routinely rotated and imaged at multiple different angles to create a 3D picture. Because rotating via mechanical microtools such as tiny grippers, needles or pipettes can damage samples, contact-free approaches such as optical tweezers, magnetic manipulation or optothermal forces are often used instead. However, these methods rely on the sample being birefringent, magnetic, light-absorbing or specially shaped.

By contrast, the laser-based technique developed by a team led by Moritz Kreysing and Fan Nan from the Institute of Biological and Chemical Systems at the Karlsruher Institut für Technologie (KIT) in Germany can rotate and spin floating microscopic objects no matter what their shape or material properties.

Their method involves focusing a 1455 nm infrared laser beam on a small region of a highly viscous liquid – initially at 25°C – in which the sample will be suspended. This gently heats that region of the liquid by a few degrees. But rather than focusing on one spot, an acousto-optic deflector steers the beam rapidly along a carefully designed two-dimensional (2D) pattern.

“Because the viscosity of the liquid changes with temperature, this moving heat pattern generates tiny fluid flows, known as thermoviscous flows. By choosing the scan geometry and timing correctly, these flows become three-dimensional and helical, like microscopic corkscrew-shaped currents,” explain Kreysing and Nan. Once a small object is floated freely in these 3D helical flows, it can be rotated, spun, transported or stabilized solely via software-defined scan patterns, with different symmetries controlling the movement of the surrounding liquid.

The KIT technique allows stacks of high-resolution images focused at different depths to be acquired along multiple different angles by re-orienting the sample, a feature that “can help overcome the anisotropic resolution of conventional 3D microscopy”, explain the researchers. Using a high-viscosity liquid to suspend the samples not only enables kinematic stop-and-go actuation, it also suppresses Brownian motion (the jiggling around of tiny particles in fluids), which otherwise decreases image resolution and makes the study of cellular transport processes tricky.

Since previous methods of laser-driving fluid flows could only create motion in one plane, the researchers found it “particularly exciting” when they first observed their laser scanning in a 2D plane generating 3D flow fields. “Another key moment was the observation of opto-hydrodynamic focusing. The particles did not just rotate; their spiral motion converged toward a stable height and position. This was significant because it meant that the optofluid could stabilize the sample, rather than requiring mechanical confinement or complex feedback control.”

As detailed in their recent Light: Science & Applications paper, the KIT team demonstrated rotation and spinning of various samples including stained biological cells, nano-printed micro-tiles and perfectly spherical beads. “We see immediate potential in advanced optical microscopy, especially multi-view imaging of suspended cells, cell clusters, organelles, microstructures or soft materials,” Kreysing and Nan tell Physics World.

Since the basic hardware components of their system will be familiar to many optics and microscopy labs, they are hoping for early adoption by specialized research groups. But to optimize the technique for biological applications, they will now investigate the use of different liquids and operating conditions “that better preserve the viability of live mammalian cells while still providing sufficient stability for precise rotation and imaging”.

The researchers suggest that future areas of application could include microfluidics, microrobotics, colloidal assembly, materials science and 3D microfabrication. As such, they also want to develop more advanced scan patterns that can generate curved, programmable or multi-axis helical flows. “This could allow more complex 3D manipulation, such as controlled orientation of irregular objects, coordinated actuation of multiple particles, or integration with microfabrication workflows,” they explain.

The post Laser-based rotation of floating samples enables high-resolution 3D microscopy appeared first on Physics World.

Obstructive sleep apnoea (OSA) is a sleep disorder in which the throat muscles relax and block the airway during sleep, causing pauses in breathing. If left untreated, OSA can lead to serious health problems. A research team led by Mingjiang Wang at the Harbin Institute of Technology, Shenzhen in China, in collaboration with Huizhou University, has now come up with a new method for detecting OSA – based on recordings of the subject snoring.

“OSA is a common yet underdiagnosed sleep disorder,” explains first author Heng Li. “The clinical standard for diagnosis is polysomnography [PSG], which is expensive, time-consuming and unsuitable for large-scale or home-based screening.”

PSG collects multiple physiological signals via various sensors that patients wear while asleep, potentially disrupting their sleep patterns. “Therefore, we aim to develop a non-contact detection method using snoring, which is a typical symptom of OSA and can be recorded contactlessly using a microphone,” says Li.

The acoustic patterns of snoring, produced by the vibrations of upper airway, contain markers that reflect airway narrowing and collapse. However, developing a reliable snoring-based detection model is challenging, as labelled snoring datasets are scarce and snoring sounds vary greatly between individuals.

In this study, published in Physiological Measurement, the research team uses a pretrained audio model called Wav2vec 2.0 to address these challenges. By transferring acoustic knowledge learned from large-scale unlabelled speech and audio data, the model removes the need for labelled snoring datasets.

Wav2vec 2.0 was originally designed for speech recognition and is computationally intensive, making it unsuitable for routine OSA monitoring. Thus the team adapted it for snoring sounds, removing layers more relevant to speech recognition while retaining acoustic features relevant to snoring pathology, aiming to lower the computational cost.

“We do not use the pretrained model directly,” Li explains. “Instead, we prune higher speech-related Transformer layers and fuse low- and mid-level representations, so that the model emphasizes snoring-related acoustic information.”

Constructing a snoring dataset

In collaboration with the Shenzhen People’s Hospital, the team collected sleep recordings from 100 individuals with apnoea-hypopnea indices (AHI) ranging from 2.4 to 68.3 events/hour. AHI – defined as the average number of apnoea (where breathing stops completely) and hypopnoea (shallow breathing) events per hour of sleep – is used to classify OSA from normal (AHI of below 5) through to severe (30 or above).

All subjects slept in hospital wards and were also monitored with a PSG device during sleeping. Sleep physicians then used the PSG data to annotate the audio recordings with apnoea and hypopnea events. Each recording was divided into 5-s audio segments and assigned as either abnormal, if they included apnoea/hypopnoea events, or normal.

To evaluate their framework’s ability to detect abnormal snoring, the researchers compared it against five other pretrained audio models. They first ran a subject-dependent evaluation, in which random audio segments from all subjects were used for training, validation and testing. All models performed well – as expected when the training and test data come from the same subjects – but the adapted Wav2vec 2.0 model achieved the highest accuracy and sensitivity.

The team then performed a subject-independent evaluation, which better reflects a real-world scenario, with each subject’s data randomly assigned to the training, validation or test set. Here, the existing methods exhibited a reduced accuracy of between 65% and 71%, while the new model retained an accuracy of 73.95%.

“In subject-independent testing, this adapted model achieved better performance than the compared baseline models, suggesting improved generalization to unseen subjects,” says Li. The team’s model was also significantly more computationally efficient, an important factor for OSA screening in home-based or resource-limited scenarios.

AHI estimation

Finally, the researchers designed a snoring-based AHI estimation framework, using their detection model to identify abnormal snoring segments and calculating a ratio-based statistic – the number of abnormal segments divided by the total number of segments – to quantify OSA severity. They then used four common linear regression models to estimate AHI for individual subjects.

All models demonstrated reasonable accuracy and good correlation between predicted and PSG-derived AHI values. The mean absolute error was roughly 11 events/h, which means that the estimated AHI may deviate from the true value by 11 events per hour of sleep and could (if near a threshold between categories) result in an incorrect severity diagnosis. However, while the current model is not yet precise enough for clinical diagnosis, it could prove ideal as a potential tool for home-based OSA screening or pre-screening.

The researchers are now investigating how to more effectively apply pretrained audio models to clinical snoring analysis. “We are focused on making these models lighter and more suitable for home-based screening, while validating their effectiveness across large, multicentre datasets,” Yun Lu, one of the corresponding authors from Huizhou University, tells Physics World. “Additionally, we plan to improve the accuracy of AHI estimation by integrating snoring-based predictions with additional non-contact physiological information.”

The post Abnormal snoring sounds help diagnose common sleep disorder appeared first on Physics World.

Researchers in Japan have succeeded in measuring the temperature inside living cells with high precision using a new class of biocompatible quantum nanosensor – something that has been difficult to do until now even. If improved, the nanosensor could be used to characterize a wide range of biological phenomena and so help in disease diagnosis, they say.

Recent years have seen the advent of a new generation of nanoscale quantum sensors that can detect the tiny magnetic fields of biological systems. Some of these sensors rely on photons and others on electrons or spin defects – typically diamond specially engineered with nitrogen–vacancy (NV) defects. This material is made by removing two carbon atoms from the diamond lattice and replacing one with a nitrogen atom. The other “hole” is left empty, thereby creating a vacancy or defect. The spin state of the defect is influenced by the local magnetic field that can be “read out” from the way it fluoresces.

While a powerful tool, and biocompatible, this type of quantum sensor does suffer from certain limits. For one, it can be structurally inhomogeneous, which affects how it detects temperature and other physical or chemical parameters inside biological cells.

A more homogenous structure

Even though the new molecular quantum nanosensor (MoQN) works in the same way as these conventional devices, it does not suffer from this problem, explain Nobuhiro Yanai of the University of Tokyo and Hitoshi Ishiwata of the National Institutes for Quantum Science and Technology (QST), who led this research effort. This is because it has a more homogenous structure and does not contain any defects. Instead, it is made by embedding molecular spin qubits, in this case fabricated from pentacene, in nanocrystals of para-terphenyl. This design makes the structure uniform on a molecular scale and preserves the quantum coherence of the spin qubits. It is then coated with Pluronic F127, which is a biocompatible surfactant.

By detecting the spin direction of the “excited triplet state” of the pentacene qubits using a technique known as optically detected magnetic resonance (OMDR), the researchers can precisely determine the temperature of the qubits’ surroundings from the OMDR peak position. When they tested their method inside the cytoplasm of cancer cells in vivo, they found that the intracellular temperature was consistently higher than the surrounding medium.

Yanai says he embarked on this study after reading about the work of Sam Bayliss’ group at the UK’s University of Glasgow, and Ashok Ajoy’s group at the University of California, Berkeley in the US on OMDR in pentacene-doped para-terphenyl crystals. He says he immediately got the idea that nanocrystals of this material could be used for quantum sensing inside cells. This was because his group had already developed such nanocrystals for a different purpose in previous research.

Ensuring biocompatibility

“I then spoke with Hitoshi Ishiwata, who is an expert in quantum sensing using NV centres,” he recalls. “While many molecular qubits have been developed to date, there had been no examples demonstrating their sensing ability within living cells.”

The project required materials science expertise, he tells Physics World, and in particular, finding out how to reduce the material to the nanoscale and ensuring it was biocompatible.

“We already knew that nanodiamonds are good quantum sensors for temperature measurements, but I had noticed a practical limitation: their ODMR spectra often vary significantly from particle to particle,” he says. “This spectral dispersion can introduce errors, especially when trying to perform precise measurements at the single-particle level.”

Replacing hydrogen with deuterium

The researchers thought they had overcome this problem during the first run of their experiments because they found that different particles showed identical OMDR spectra. However, their joy quickly waned when they observed that the spectra were still broadened by hyperfine interactions between the pentacene-doped para-terphenyl molecules’ electron spins and hydrogen nuclear spins.

To improve the spectral resolution, Ishiwata says he suggested chemically modifying the molecule by replacing the hydrogen in it with deuterium. And the technique worked: “the hyperfine broadening was strongly suppressed, allowing us to determine the OMDR spectra much more precisely.”

These findings, which are detailed in Science Advances, show that MoQNs are a chemically versatile platform for quantum sensing in living cells and that they can operate directly inside them while maintaining the precision needed for absolute thermometry, he says. Their appeal also lies in in the fact that their structures can be easily modified.

It will not all be plain sailing, however, adds Yanai. MoQNs cannot yet target specific organelles within cells, so endowing them with this targeting capability is an important future challenge. “What is more, their size has been limited to around 200 nm so far, so creating smaller MoQN particles will be crucial,” he says.

The post Molecular spin sensor takes the temperature of cancer cells appeared first on Physics World.

Google upgraded iPhone migration options with a service built directly into Android 17 aimed at simplifying the switch, a release timed to coincide with the unveiling of Samsung’s Z Series at its Galaxy Unpacked event.

The search giant expanded the range of data which can be wirelessly transferred from an iPhone without downloading a separate migration app.

In addition to “photos, videos, contacts, messages and calendar”, the latest method covers a user’s “Google Account, passwords, Wi-Fi credentials” and eSIM profile.

By simplifying onboarding and broadening the compatible data types, Google aims to lower the barriers to switching smartphone ecosystems to attract more iPhone users to Android.

Google commenced deployment of the feature on select Pixel devices and it is also available on the Samsung Galaxy Z Flip8 and Z Fold8 ranges unveiled at Galaxy Unpacked.

The company plans to expand the migration method to more Android devices in future.

The post Google upgrades iPhone switching tool appeared first on Mobile World Live.

Samsung Electronics expanded its foldables lineup with a trio of smartphones at its Galaxy Unpacked event in London, including the first Ultra-branded foldable as it pushed upgraded form factors and AI-powered features.

Leading the charge was the Galaxy Z Fold8 Ultra, the first device to carry Samsung’s Ultra branding on a foldable. Measuring 4.1mm when unfolded, Samsung said the device is its thinnest Galaxy Fold model to date.

Powered by Qualcomm’s Snapdragon 8 Elite Gen 5 Mobile Platform for Galaxy, the handset features an 8-inch main display, 5,000mAh battery and a 200MP main camera. It also introduced new Galaxy AI tools including Gemini Notebook, capable of summarising documents and generating content from uploaded files, and Now Nudge, a contextual Galaxy AI feature designed to proactively prompt actions based on user activity.

Samsung also refreshed the Galaxy Z Fold8 with a redesigned display format and adjustable aspect ratios aimed at improving content consumption across video, gaming and reading. The device weighs 201 grams, making it the company’s lightest Fold model, and includes a 4,800mAh battery rated for more than 26 hours of video playback. The Fold8 also introduced an AI-powered editing tool to automatically track and reframe a subject in a video.

In the clamshell segment, Samsung updated the Galaxy Z Flip8 with expanded functionality on its FlexWindow cover display, enabling users to access apps, messages and personalised AI information without opening the handset. The latest Flip model also adds a 50MP main camera and dual recording mode for capturing both subject and creator simultaneously.

Pricing starts at $2,099 for the Galaxy Z Fold8 Ultra, $1,899 for the Galaxy Z Fold8 and $1,199 for the Galaxy Z Flip8. All three models will be available from 7 August.

At the event, TM Roh, CEO, president and head of Samsung Electronics’ Device eXperience division, emphasised the importance of foldables in the AI era.

“The question is: if AI is becoming more personal, more proactive, and more multidimensional, what is the optimal device to bring that experience to life? We believe the answer is foldables.”

Wearables
Samsung also refreshed its wearables portfolio with the Galaxy Watch Ultra2 and Galaxy Watch9, its first watches built on Qualcomm’s Snapdragon Wear Elite platform.

The smartwatches introduce new AI-powered health features, with the vendor claiming both models received an FDA-cleared sleep apnea detection update. The Galaxy Watch9 starts at $379, while the Watch Ultra2 is priced from $699.

In addition, Samsung unveiled intelligent eyewear, although pricing and a release date were not disclosed.

Doubling down
Ben Wood, chief analyst at CCS Insight, said Samsung is “doubling down on foldables at exactly the moment the category is about to get its biggest credibility boost with the much-rumoured arrival of a folding iPhone later in the year.”

He said Apple’s expected entry “could also be the moment that turns foldables from a niche premium category into a mainstream flagship battleground.”

Wood added “Qualcomm is one of the quiet winners from Samsung’s latest launch, securing important design wins across phones, watches and glasses.”

The post Samsung unveils 3 new foldables packed with AI appeared first on Mobile World Live.

iOS 27 Developer Beta 1 has been out since WWDC 2026, with the second beta slated to be released soon. There are a number of changes and performance improvements to be appreciated in this very first developer beta, along with oddities to be expected with prerelease software. Just like last year, I decided to take […]

The post iOS 27 Beta 1: Impressions on Build 24A5355q after 1 week appeared first on Low End Mac.

On June 8th 2026 at WWDC 2026, Apple announced the iOS 27 Developer Beta, which is set for consumer release in September 2026. iOS 27 was announced alongside macOS Golden Gate, iPadOS 27, visionOS 27, watchOS 27, and tvOS 27. Aimed to be a like a “Snow Leopard” sort of release, this new version brings […]

The post iOS 27 appeared first on Low End Mac.

Consumer hardware company Aina raised millions of dollars in seed funding to finance development of an interface it argues is required to fully unlock the potential of AI on smartphones and PCs.

The company believes current device hardware lags the progress of software and AI, resulting in a disconnect between expectation and reality in terms of the interface with apps and services.

It raised $5.5 million in the recent funding round to advance development of experimental interfaces which adapt to the app being used.

Aina stated it already shipped hundreds of experimental keypads called Dune to early adopters, a product for Mac computers it describes as “context-aware” which primarily focuses on work tools.

It is using the feedback from Dune to develop a general-purpose interface for smartphones and computers.

Aina founder and CEO Apoorv Shankar (pictured) said current smartphones and PCs “are still primarily designed for browsing” and use manual inputs requiring users to “figure out what you need”.

He argued this is laborious when users are “doing hundreds of tasks” per day.

The company believes there is room within the broader AI assistant space for a contextualisation system which makes it easier for people to interact with the technology. Shankar explained the general-purpose interface on the drawing board is being designed to “capture human approval”.

Shankar was previously VP of hardware at smart ring company Ultrahuman.

The seed funding round was led by Redstart Labs (Infoedge) and 360 ONE Asset, with two other investment companies and a host of angel investors.

Along with launching its interface, Aina intends to use the backing to grow its team in the US and India.

The post Aina scores $5.5M for AI interface work appeared first on Mobile World Live.