Apple’s Spatial Reframing tool in Photos for iOS 27 is an interesting use of Apple Intelligence, but don’t push it too far just yet.

WWDC 2025 included a neat feature for Spatial Photos that lets users convert flat images into 3D scenes. By moving the iPhone around, you could temporarily re-angle your shot and explore the scene.

It was a neat trick, but it wasn’t that much of a major feature for photographers. It did, at least, give you an idea of how a Spatial Photo would look on something like an Apple Vision Pro.

One year later, Apple has decided to try and get the feature to be more practical to iPhone users. That comes in the form of Spatial Reframing.

Fixing the imperfect shot

One of the problems with photography is regretting not lining up the shot perfectly. Frequently, you’ll look back at what you’ve just taken, and the background isn’t in quite the position you wanted it to be.

If you were to edit it traditionally, you would try to cut around the subject and move it over, then fill in the new empty pixels with a clone tool or something else. An expert image editor can do this, and most people won’t tell that an edit has taken place at all.

That requires time and skill, which the average photo-taker doesn’t really have or wish to invest.

Spatially reframing an image in iOS 27

Advertisement

Spatial Reframing is a blend of the previous Spatial Photos feature and generative AI smarts. The idea is that you can select an image, the iPhone will analyze it, and then you can alter the angle of the camera’s view to a new one.

In theory, that would be a quick and relatively painless process, and with no issues at all. Depending on the photo you throw at it, you may just get that, but with some massive caveats.

Background filling

You can find the Reframe feature under the editing section of an image, under Tools.

Once you tap it, the screen fills up with a multi-colored filter, as it scans the shot. Once scanned, you’re instructed to touch and drag to adjust the perspective.

You can also use a two-finger pinch to pan, zoom, and rotate the image.

Dragging the picture around gives you a similar effect to the Spatial Photos, but to a more extreme degree. You can move the angle far enough to one side or the other that you can uncover sections of the background that simply aren’t visible in the original shot.

Reframing a portrait of a kitten works well. Original on the left.

In the preview, this is filled in with a minimal generative graphic that won’t be used in the final picture. This is especially the case for the edges, which appear blurred because it’s a lot to generate on the fly, and that’s not necessary in a preview.

Once you have set your new angle, hitting the Reframe button sets the processing in motion. After a few seconds, you have your reframed image.

In the short time we have played with the feature, we tried it out with a pair of images, to see how it works with a close-up portrait and with a wider scene.

The portrait, which we used an old image of a kitten, was handled pretty well. It was a minor shift of the camera to the right, with a small amount of deformity to the cat’s image.

More apparent is the background, as the left-hand side of the image was completely generated by the tool. It did pretty well in its blurry state, and if I showed anyone who didn’t know the room’s layout, they wouldn’t tell.

Our second test was with a wide touristy photograph of the Colosseum in Rome. It’s a difficult structure with many archways, and the subjects are a distance away from the camera in the middle of the frame.

When lining up the shot, we could tell that it was trying to make a vague-but-acceptable background for the preview, which is fine.

The final image has some plusses, but also some minuses.

Reframing a tourist scene initially looks OK. Original on the left.

Advertisement

On the plus side, it handed generating the background really well. Arches and road that were not visible in the original shot were created and put into place in the back quite well.

Less well done are the subject faces. You can tell that, as part of the reframing, the bodies and heads are taken into account, and are similarly adjusted to match the rest of the image.

This can sometimes work well, but the resulting warp to the faces is unflattering, to put it mildly.

Caveat Emptor

Spatial Reframing, as a concept, makes perfect sense. If you have a camera system and processing that can take apart a scene, move elements around, and smartly generate missing bits, there’s no reason not to do it.

This would be a massive task for a human to undertake, so what it’s coming up with is pretty phenomenal for a first try.

Close-ups of the original [left] and the warped faces of the reframe [right]

That said, we are talking about a feature that is in a developer beta, that is months from release, and the first real attempt too. It’s expected that there will be hiccups and foibles here.

Expect more improvements in the future.

As it stands, it’s a nice feature that could make for some fun shot changes. Content altered by the feature is not going to make the cover of Vogue anytime soon, so professional editors can breathe a sigh of relief.

If you don’t push it too far, it’s decent enough to make your Instagram cat photos a bit better.

Open Media: After years of development, an industry consortium has published the first major release of AV2. The next-generation video encoding standard has ambitious goals, including improved compression efficiency and broader industry adoption, while remaining royalty-free.

The Alliance for Open Media (AOMedia) recently released version 1.0.0 of the AV2 specification and reference code. The finalized AV2 specification provides a baseline framework for software and hardware developers, who can now begin integrating the new video codec into products across the media and technology industries.

AOMedia introduced AV2 as a next-generation video coding standard built on the same foundations as AV1. Both codecs are royalty-free, open-source solutions for streaming, transcoding, and other media-related workloads, offering an attractive alternative to royalty-bearing formats such as AVC/H.264, HEVC/H.265, and the newer Versatile Video Coding (VVC) standard.

The final AV1 specification was released in 2018, while AV2 took longer than initially anticipated to reach completion. AV1 has since gained significant traction across the industry, with Netflix and other major players adopting the royalty-free codec at an accelerating pace. AV2 was originally expected to arrive in 2025 and promises substantial improvements over its predecessor in image quality, feature support, and compression efficiency.

Early testing suggests that AV2 can deliver roughly 30% better compression efficiency than AV1, while significantly outperforming older codecs such as VP9 and H.264. According to AOMedia, AV2 has been engineered to provide “superior” compression efficiency compared to AV1, enabling media companies to deliver high-quality video streams at substantially lower bitrates. Alternatively, they can maintain similar bitrates while offering higher video quality.

AV2 was designed to meet the evolving needs of the video industry, including streaming, broadcasting, and real-time video conferencing. The new format also introduces improved support for mixed-reality applications, split-screen delivery of multiple video streams, and a broader range of visual quality options.

AOMedia was founded in 2015 by major technology companies to develop open video standards capable of succeeding Google’s VP9 codec. The non-profit consortium includes industry giants such as Amazon, Google, Intel, Nvidia, and Microsoft, alongside smaller technology organizations such as Mozilla.

AOMedia created AV1 to reduce the substantial licensing costs associated with patented video standards such as AVC and HEVC, and the strategy appears to be paying off. Modern graphics processors from Nvidia and AMD fully support hardware-accelerated AV1 encoding and decoding, alongside older formats, while support for VVC remains largely absent despite the codec having been available since 2020.

In contrast, VVC’s adoption has been hindered by a complex licensing landscape, leading some industry observers to question its long-term prospects. While it may take years for AV2 support to arrive in GPUs and other hardware devices, software adoption is already moving forward. The VideoLAN community has released dav2d, its portable AV2 decoder, featuring architecture-specific optimizations for x86 (AVX2), ARM (AArch64 NEON), and RISC-V processors.

macOS Tahoe threw an icon on every menu item, making them impossible to distinguish at a glance. macOS Golden Gate has rectified that design taboo with blessedly iconless menus.

In Disney’s 2004 animated film The Incredibles, a very basic concept of “when everyone is special, no one is” is explored. Alan Dye must not have seen that movie or the many like it.

Well, now that he’s gone, so is his team’s decision to frustratingly place icons next to every menu item. Apple announced macOS Golden Gate on Monday, and it has fixed this problem.

macOS Tahoe introduced the controversial Liquid Glass design, but that may not have been the most hated change to the platform. Oh, the weird Finder icon was a problem Apple fixed immediately, but I’m talking about menu icons.

For some reason, Apple went against its own age-old design guidelines to add special little SF symbols to every menu item. Every single item in the menu got these little pictures of cogs, squares, and pencils.

The problem is, when every menu item has a little picture next to it, your brain stops distinguishing between them. They’re treated as another letter in a line of text and may as well not exist.

Well, someone got the memo because macOS Golden Gate ditches the icons again. Here’s a side-by-side of the new menu next to the previous one.

The left menu shows a lack of icons in macOS Golden Gate and the right menu shows the icons in macOS Tahoe

Advertisement

Apple spent a lot of its keynote addressing user complaints. Even the corner radii being mismatched was addressed directly.

WWDC 2026 is underway and AppleInsider is digging through all of the betas for every small change. Even though Apple spent most of its keynote address talking about Apple Intelligence, there are many new features and updates across every operating system.

For years, Apple has been accused of being one of the biggest stragglers in the AI arms race. Doubters have argued that Apple’s lack of a clear AI strategy have cost it its edge, and Wall Street analysts have worried that the gap could start hurting iPhone sales.

Now, the company has unveiled what it is billing as its biggest AI launch to date: Siri AI, which embeds new automated capabilities (fueled by a partnership with Google Gemini) into the very spine of its software.

Is it enough to get people to stop saying that Apple is “losing” the AI race?

To be honest, nobody really knows. But the question itself may be the wrong one. A better one might be: are Apple customers actually going to use these features and, if they do, will it help Apple’s business?

Before we address that question, we should note that Monday’s announcements also came with an interesting comment from Craig Federighi, Apple’s senior vice president of software engineering.

“Some appear to be racing forward, seemingly pursuing AI for the sake of AI, without clear regard for the people — all of us — that it’s ultimately meant to serve,” Federighi said during his remarks. “At Apple, our mission has always been to turn the potential of advanced technology into helpful and intuitive products for everyone.”

The not-so-veiled defiance on display here seems like both a response to Apple’s “behind-on-AI” criticism and an effort to acknowledge the deeply ambivalent — and, according to some polls, increasingly negative — sentiments that many consumers have about the AI industry. It’s also a shrewd message at a moment when Americans are worried that AI will take their jobs and rot their brains. Apple is positioning itself as the AI company that’s actually on your side.

Judging by Monday’s demos, that positioning has some substance behind it. Siri can now surface information buried deep in your inbox or text history and surface helpful information and offer helpful suggestions based on it. It can use what Apple calls onscreen awareness to give you context about what you’re looking at. And — using Gemini — it can pull near-instantaneous up-to-date information from the web and deliver it right to your device.

Siri is also designed to work seamlessly across Apple devices, giving users increased flexibility and, like other AI chatbots, it stores chat histories so users can revisit past conversations.

By building AI functionalities into its disembodied, ethereal assistant, Apple also has the potential to eat into the advantages of competitors whose apps can only reach users through its own App Store. For those competitors, having Apple’s AI embedded at the operating system level is a meaningful threat to their distribution advantage.

The keyword here is “potential” since this version of Siri won’t be available to consumers until later this year, as a beta.

A final verdict will have to wait, but what’s already clear is that Apple is doing its best to court its audience — whether they end up going for it or not. Apple is obviously a hardware company, and these updates are designed to make that hardware incrementally more user-friendly and convenient, keeping users glued to their devices a little while longer.

The contrast with its competitors is instructive and maybe the most important signal in Monday’s announcements for anyone watching where the AI industry is actually headed. Take OpenAI, which, despite shipping updates at a relentless pace, has struggled to define who it’s actually selling to, oscillating between consumers and enterprises. Or Meta, which is pouring gargantuan sums into AI without a clear explanation of how it connects to the company’s core advertising business.

Apple’s more measured approach is starting to look optimal by comparison — and more financially sound. For the most part, Apple hasn’t needed a gangbusters AI strategy. It posted historic iPhone sales last quarter. And as questions mount over AI’s profitability and real-world utility, Apple is spending significantly less than its competitors — roughly $14 billion in capex planned this year, against a cumulative $900 billion being committed by other tech giants — while still earning huge amounts of revenue. That revenue has come from the AI industry itself via taxes on AI companies that use its App Store to platform their apps.

In short, Apple is spending less, making more, and now launched a suite of AI features that — for many iPhone users — will feel indistinguishable from the other AI applications already available to them through the App Store. If that doesn’t exactly count as “winning the AI race,” it may be the smartest way to run it.

Instagram head Adam Mosseri announced the feature a year ago.

New variants of the NFCShare Android malware are being distributed as fake updates for legitimate banking apps hosted on GitHub.

The malware has evolved and is now targeting customers of multiple banks and financial institutions across Europe in a phishing campaign aimed at stealing payment card data.

After tricking victims with a fake verification screen to place the cards near the mobile device’s near-field communication (NFC) chip, NFCShare reads the information using Android’s IsoDep interface and EMV commands.

The malware steals the card number, type, expiry date, and a 4-digit PIN entered by the victim under the pretense of a security step, and exfiltrates it to the attacker’s command-and-control (C2) host over a WebSocket channel.

The information collected this way can then be used in NFC payment relay schemes, as documented in the NGate, SuperCard X, and RelayNFC malware attacks.

NFCShare was first documented by D3Lab researchers in January 2026, who have been tracking its activity and evolution.

D3Lab researcher Andrea Draghetti told BleepingComputer that, despite similarities to other Android malware that exploit NFC chips for data theft, NFCShare uses distinct code, libraries, architecture, and implementation details.

Draghetti noted, though, that it could still be an evolution of the same ecosystem, driven by the same threat actors.

Recent NFCShare attacks observed starting May 14 begin with the victim visiting a phishing site that impersonates a real bank and asks for banking credentials.

Victims are then urged to update their banking app and are redirected to a GitHub repository hosting a malicious APK file.

The researchers note that SMS messages or phone calls from fake bank representatives may also be used as part of the social-engineering process, as seen in similar attacks, although D3Lab researchers did not observe these methods directly.

Since its creation on April 10, the GitHub repository used for distributing NFCShare has hosted 56 unique APKs that impersonated mobile apps for banks primarily from Italy and Spain:

• Intesa Carte.apk
• Sella Carte.apk
• Banca Sella Carte.apk
• Nexi Carte.apk
• Fideuram Carte.apk
• Mooney Carte.apk
• CaixaBank.apk
• CaixaBankNfc.apk
• CaixaReactivaTarjeta.apk

In January, D3Lab reported that the malware targeted only Deutsche Bank in Germany, which may suggest an extended targeting scope.

One interesting aspect of the new version of the malware is the introduction of malformed APK packaging to hinder automated analysis, and potentially also security tools.

The APK is still a ZIP archive, but the newer samples include poisoned/malformed file paths within that ZIP, causing some extraction tools to wrongly interpret internal relative paths as filesystem paths and trigger errors.

However, D3Lab notes that this trick does not prevent manual analysis or code recovery; rather, it disrupts static analysis in certain tools.

Android users are advised to source banking apps only from Google Play, enable Play Protect, and be cautious of “verification requests” that prompt NFC card scans.

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

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

Get the whitepaper

EOFY sales in Australia aren’t just about work-related deals — nothing is stopping you from picking up something for your home or yourself when there are savings to be had.

So if you’ve been eyeing something from Dyson, you’re in luck because the British brand is hosting its own EOFY sale that’s slashing up to AU$651 on some of its top-selling products.

You don’t have to buy directly from Dyson if you don’t want to, though. Other authorised retailers are also discounting Dyson’s tech and I’ve hunting down some of the best prices on a variety of premium vacuums and hair stylers. I’m also keeping an eye on Dyson’s purifier fan heaters for good discounts, so watch this space.

I’ll be tracking all the best Dyson EOFY deals through June, so you can bookmark this page and check back to see if there have been any price drops or new items discounted — you’ll get the most up to date information right here.

Top Dyson retailers

Best EOFY 2026 Dyson deals

Vacuums

Fans, heaters and air purifiers

Other Dyson deals

AI + ML

L’etat, c’est AI

OPINION The US government is reportedly weighing whether to take a financial stake in AI companies, which looks a bit like negotiating for a seat on the Titanic.

Neither OpenAI nor Anthropic, the marquee brands in US AI, are profitable yet. While Anthropic may be nearer to that point if its accounting survives scrutiny, OpenAI’s $1.4 trillion in financial commitments over the next eight years have been interpreted as a red flag for investors.

This raises (at least) two questions: Should the US government be picking winners? And should the US government be picking losers?

The first question appears already to have been decided. As noted by the US Council on Foreign Relations, since January 2025, the feds have invested $20.9 billion in sixteen deals that involve direct ownership. This represents a change from more hands-off financial arrangements involving grants, loans, and tax incentives.

The Department of Commerce, for example, has taken a 10 percent stake in Intel, once a symbol of American technical prowess and now a national security backstop. The Development Finance Corporation had invested in minerals, energy, and infrastructure. And the Department of Defense has undertaken at least seven similar deals.

Neoliberal US notions about competition and the separation of church, state, and private industry have succumbed to the new world disorder.

When economists from Harvard and Yale looked at the issue in a 2021 paper titled “The Dance Between Government and Private Investors: Public Entrepreneurial Finance around the Globe,” they were cautiously optimistic.

Authors Jessica Bai (Harvard), Shai Bernstein (Harvard), Abhishek Dev (Yale) and Josh Lerner (Harvard) looked at 755 entrepreneurial finance policies in 66 countries during the period from 1995 to 2019. They concluded that “government funding programs are associated with subsequent increases in innovation,” as measured by “top patents.”

They offered some caveats, such as the observation that “government programs frequently rely on private capital markets through capital matching requirements, where private capital groups are often allowed to invest in more preferential terms than the public funds.”

And they also noted that economists have long recommended government investment in response to market failures – areas where private funding has chosen not to invest, presumably due to the uncertainty of returns.

A recent example of that would be the US Commerce Department’s decision to invest $2 billion in quantum computing in exchange for a minority controlling stake in nine technology companies. Pure-play quantum computing companies like D-Wave, Quantinuum, IonQ, and Rigetti Computing are not making a profit. But concern that quantum computing might some day do meaningful computing not possible with classical computers is enough to keep the funds flowing for now.

The US government’s reported interest in AI companies might be interpreted in a similar light, as a bailout for companies that have committed to spend heavily on data centers before demand has been demonstrated and pricing has stabilized. With OpenAI and Anthropic preparing to go public, the White House would do better to wait before placing its bet.

OpenAI CEO Sam Altman is said to have pushed for federal investment last year but publicly repudiated the idea after CFO Sarah Friar suggested federal loan guarantees. If the feds were to buy into OpenAI, the deal might take the form of a public wealth fund – so the public would receive revenue from intellectual property that AI firms have captured and are reselling.

US Senator Bernie Sanders (I-VT) last week said he planned to introduce a bill called the American AI Sovereign Wealth Fund Act. Funded by a one-time 50 percent tax paid in AI company stock, it would give the public a say in how AI is used and a portion of the revenue generated by AI companies (which, again, follows from the largely uncompensated capture of public content).

Meanwhile, the White House last week issued an executive order directing “the national security enterprise to accelerate AI adoption to meet surging demand, adapt the best commercial and open-source technologies for mission use, assure that fielded systems are robust, steerable, controllable, and preserve clear lines of accountability under the Constitutional chain of command.” And the order promises “new partnerships with willing private-sector companies to secure America’s cutting-edge AI against global threats.”

Buying into these companies doesn’t make a lot of sense if they can deliver on their promises at a viable price. The market would ensure plenty of good options for federal procurement.

But if leading AI models are priced like Claude Mythos, reported to run $25 per million input tokens and $125 per million output tokens, or about 5x Opus 4.8, there may be some concern that leading edge AI will be too costly for much of the market. Uber’s $1,500 monthly token spending cap per employee AI tool suggests companies won’t reward the AI industry for over-investing. If cutting-edge AI is going to be priced out of reach for most industries and if it really can accomplish things that lesser models cannot, the case for federal involvement gets stronger.

It would be a shame if the feds rewarded OpenAI and its peers with taxpayer money because that would reward fiscal irresponsibility and hinder startups hoping to innovate. Worse still, it would commit funds prematurely and unnecessarily for some notional national security edge that’s razor thin and is being dulled by evolving open weight models and foreign model providers. 

The jury is still out – there are at least 115 lawsuits against AI companies – on whether there’s a broad, sustainable market for AI services outside of software development and perhaps a few other knowledge work markets. The government should wait for the courts, the public, and the market to weigh in before riding to the rescue. ®