From Dominic Preston at The Verge:

But unlike Samsung’s previous squarish Folds, which weren’t much wider than simply turning your phone on its side, opening the Z Fold 8 up really does fit more video onto your screen, and the wider aspect ratio will be much more natural for gaming too. It feels less like a foldable and more like a tablet and is all the better for it.

There's a misconception that Apple arrives 'fashionably late' to parties and stars all the limelight after its competitors have put in all the effort working the room.

The reality is closer to this - Apple lets its competitors do their R&D in public, and on customers who suffer through many iterations of mediocrity, before either the technology or just plain sensibilities are in place to make a device it can recommend to everyone. Apple is happy to, and has the resources to, do its testing behind closed doors and only release something to market when it knows with some level of conviction that . This isn't to say that their v1 product releases are the finished article, or that it's a faultless strategy, but that to a large degree the foldable iPhone will be a 'ready for prime time player' rather than a prototype.

Which is what makes the foldables market, if Dominic Preston's take is accurate, so amusing - Samsung have had a seven year head start on Apple to figure out the perfect form factor for a folding device and have iterated on pretty much the same 'ultra tall' folded aspect ratio every time. But then we all caught wind that Apple's soon-to-be-released contribution will settle on a much squatter silhouette - one which makes for a more jarring appearance when folded but feels like a much more natural mini-tablet when folded out.

Samsung, having now finally tried this in one of its own secret labs, has now realised that of the two prototypes it gave the market to decide between (namely the Z fold and Z flip), neither were winners, and that the winning candidate was actually prototype #2512 that's sat in a drawer with wacky 3d-printed hinge mechanisms from 2019.

Sam Altman at OpenAI's blog:

Last week, Hugging Face disclosed a new kind of security incident after they detected and contained an AI agent that compromised their infrastructure, something we expect to become more commonplace with the proliferation of increasingly cyber-capable models. After investigating, we now know that this particular incident was driven by a combination of OpenAI models [...]

That this happened eventually its not a shock. Part of the training process, as per his post, is to take the stabilisers off the bike and goad it into doing its worst:

This incident occurred during an internal evaluation which prompts models to pursue advanced exploitation using complex attack paths, in an effort to quantify their cyber capabilities. We estimate maximal cyber capabilities by running this evaluation without production classifiers used to prevent models from pursuing high-risk cyber activity.

What's more frightening is the artificial constraints put on the model during evaluation - namely, finding a way to grant itself internet access:

While operating in our sandboxed testing environment, our models spent a substantial amount of inference compute finding a way to obtain open Internet access [...] after gaining Internet access, the models inferred that Hugging Face potentially hosted models, datasets and solutions for ExploitGym [...] In one example, the model chained together multiple attack vectors, including using stolen credentials and zero-day vulnerabilities to find a remote code execution path on the Hugging Face servers

In the space of just a few months we've gone from the 'dangerously good at finding undocumented security exploits' Mythos moment to the 'actively escaping Alcatraz' milestone above.

Chinese models now make up the majority of model usage on OpenRouter. I sure hope those working on Kimi know both how and when to unplug the router when the moment comes.

Further reading:

When China Gets Its Own Mythos

Preparing for an AI cyber crisis.

Foreign Affairs MagazineMichael Sulmeyer

From The Verge :

While nothing has been confirmed, Microsoft’s recent moves make selling off the brand seem like a possible option, as New York University professor Joost van Dreunen tells The Verge. “A wholesale divestiture of Xbox remains on the table, and it looks likelier given Xbox’s struggles with rising hardware costs and Microsoft’s focus on AI and infrastructure,”

Of the options presented, Amazon seems like the most natural fit for some obvious reasons:

• Has attempted (and mostly failed) at building a cloud gaming service already
• Has a popular subscription base content model it can tack Xbox Live membership onto
• Has an install base of a gazillion fire sticks that could function as streaming sticks
• Loves shoving its own content top, front and centre of said streaming sticks
• Has built a business model predicated around selling hardware at or below cost in return for content revenue

It would of course be deliciously ironic to see Xcloud (are we still calling it that?) running on Amazon’s servers.

Per Mark Gurman:

Apple’s decision to use the M5 line before jumping to a future M7 is part of a broader shift in its chip strategy. The company is skipping M6-generation Pro and Max chips altogether, breaking from its usual approach. Bloomberg News reported on this strategy earlier this week.

Debate about the demand for touchscreen Macs aside, I'm not sure I can think of another occasion where Apple has introduced a new chassis design and kept the same internals. For the migrations to Apple Silicon and Intel CPUs they kept some of the previous hardware designs, presumably to allow them to focus on getting the architecture to work in more familiar territory, but presumably they're confident enough that OLED touchscreens and presumably other feature upgrades (5g modems anyone) will be alluring enough to power users that they can put up with a painfully slow and outdated chip powering it all.

unprecedented cross-product-line price hikes yesterday, other than how much the iPhone 18 line will cost when it arrives in September, is whether this is a temporary measure or a permanent one.

9to5 Mac's poll on the matter amusingly offers readers the option of responding with absolute or reasonable certainty, as if it's readers fall under either the 'friends and family of leaky-executive-team members' or 'regular Apple followers'. In any case, at the time of writing ~75% of respondents think these prices are the new normal:

Mark Gurman, who often misconstrues his own excellent track record of scoop journalism with a mix of very pedestrian opinions, seems equally convinced that $2,499+ for a Pro chip in a MacBook is now the lowest we'll ever pay for it:

I don't know that I'll be able to find a predictions market that offers me a 1million year resolution window for proving this wrong, but I'd be very surprised if Apple locks these new price tiers in for good if, and it's a fucking big if, RAM and storage prices ever return to something close to their previous rates.

For one, the Apple of recent years has been very doggedly perusing lower price points:
- The $599 Mac mini, introduced in 2020 and received a 16gb RAM spec bump in 2024
- The Apple Watch SE3, whist not a price reduction, brought enough higher-end features down to the $349 price point to make it the de-facto Watch option for most people
- The $599 ($499 edu) MacBook Neo was at the bottom end of predictions for their new line

None of this is to say that Apple has decided to run itself as a charity. The company famously has very healthy profit margin expectations for each SKU that it sells, and the fact that they've been able to make an iPad for $349 and MacBook-Air-if-you-squint-laptop for $599 is a testament to the advances in their engineering and supply chain prowess that they can do so with these margins intact.

The Neo is an especially useful weathervane for Apple's pricing strategy as it is one of the few areas where Apple is aggressively competing on price, in this case with Chromebooks, rather than being able to use its superior Operating System as leverage. If component prices allow half-decent Chromebooks to return to sub $500 prices again then Apple will surely need to follow suite to tempt educational IT departments away, and consequently will need to bring the MacBook Air price down with it to avoid leaving too big a pricing delta between their 'good' and 'better' products.

Apple's own statement on the matter seems to confirm the fact that they are, at the moment at least, uncomfortable with these prices:

The consumer electronics industry is facing an unprecedented challenge. The rapid expansion of AI data centers has created an extraordinary surge in demand for memory and storage. We have never seen a component price increase this much, this quickly. We have shielded our customers from these increases so far, but we have now reached a point where we need to begin raising prices on a number of products, including today's increases for iPad and Mac. We know this is not welcome news, and we are working tirelessly to find solutions.

Whether consumers react with similar hostility will surely determine the direction they take should the opportunity of a price-drop ever present itself.