Apple’s macOS updates have been so dull lately, the most interesting part of last year’s macOS Sonoma ended up being widgets. Widgets! Thankfully, macOS Sequoia has a lot more going on — or at least it will, once Apple Intelligence rolls out over the next few months. For now, though, Sequoia delivers a few helpful features like iPhone Mirroring, a full-fledged Passwords app and automatic transcription in the Notes app. At the very least, it’s got a lot more going on than widgets.

iPhone mirroring changes everything for Macs

Heading into WWDC earlier this year, I was hoping that Apple would let Vision Pro users mirror their iPhones just as easily as they can mirror their Macs. Well, we didn’t get that, but iPhone Mirroring on macOS Sequoia is close to what I’d want on the Vision Pro. Once you’ve got a Mac (with an Apple Silicon chip, or one of the last Intel models with a T2 security chip) running the new OS, as well as an iPhone running iOS 18, you can easily pair the two using the iPhone Mirroring app.

Once that connection is made, you’ll see a complete replication of your phone within the app. It took me a few minutes to get used to navigating iOS with a trackpad and keyboard (there are a few new hotkeys worth learning), but once I did, I had no trouble opening my usual iPhone apps and games. If you’re spoiled by the 120Hz ProMotion screen from an iPhone Pro, you’ll notice that the mirrored connection doesn’t look nearly as smooth, but from my testing it held a steady 60fps throughout games and videos. I didn’t notice any annoying audio or video lag either.

While it’s nice to be able to launch my iPhone from my Mac, I was surprised at what ended up being the most useful aspect of this feature: Notifications. Once you’ve connected your phone, its alerts pop up in your Mac’s Notification Center, and it takes just one click to launch the app it’s tied to. That’s useful for alerts from Instagram, DoorDash and other popular apps that have no real Mac options, aside from launching their websites in a browser.

iPhone Mirroring is also a sneaky way to get in a few rounds of Vampire Survivors during interminably long meetings or classes. (Not that I would ever do such a thing.) While many mobile games have made their way over to the Mac App Store, there are still thousands that haven’t, so it’s nice to have a way to access them on a larger screen. Not every game works well on Macs — it’s just tough to replicate a handheld touchscreen experience with a large trackpad — but mirroring is a decent option for slower-paced titles. I didn’t encounter any strange framerate or lagging issues, and sound carried over flawlessly as well.

I typically always have my phone within reach, even when I’m working at a desk. But picking it up would inevitably disrupt my workflow — it’s just far too easy to get a notification and find yourself scrolling TikTok or Instagram, with no memory of how you got there. With iPhone Mirroring, I can just keep on working on my Mac without missing any updates from my phone. It’s also been useful when my iPhone is connected to a wireless charger and I desperately need more power before I run out the house.

If you’re the sort of person who leaves your phone around your home, I’d bet mirroring would also be helpful. The feature requires having both Bluetooth and Wi-Fi turned on, and the connection range is around 50 feet, or what I’d expect from Bluetooth. Thick walls and other obstructions can also reduce that range significantly. In my testing, I could leave my iPhone in my backyard and still be able to mirror it in my living room 40 feet away. Naturally, the further you get, the choppier the experience.

Sure, Apple isn’t the first company to bring smartphone mirroring to PCs. Samsung and other Android phone makers have been offering it for years, and Microsoft also has the “Phone Link” app (formerly Your Phone) for mirroring and file syncing. But those implementations differ dramatically depending on the smartphone you’re using, they don’t seamlessly integrate notifications and simply put, they would often fail to connect. Once you set up iPhone Mirroring, getting into your phone takes just a few seconds. It just works. And after testing the feature for weeks, I haven’t run into any major connection issues.

Better window tiling, finally!

It’s 2024 and Apple has finally made it easier to position Mac windows around your monitor. Now you can drag apps to the sides or corners of your screen, and they’ll automatically adjust themselves. It’s allowed me to quickly place a browser I’m using for research alongside an Evernote window or Google Doc. Similar to Stage Manager in macOS Ventura, the tiling shortcuts are a significant shift for Mac window management.

And, of course, they’re also clearly similar to Windows 10 and 11’s snapping feature. Given that much of Apple’s UI focus is on iOS, iPadOS and VisionOS these days, it’s easy to feel like the Mac has been left behind a bit. I don’t blame Apple for cribbing Microsoft’s UI innovations, especially when it makes life easier for Mac users.

Slick video conferencing background replacement

Apple has offered lighting adjustments and portrait background blurring in video chats for years, and now it’s using that same machine learning technology to completely replace your backgrounds. Admittedly, this isn’t a very new or exciting feature. But it’s worth highlighting because it works across every video chat app on your Mac, and since it’s relying on Apple’s Neural Engine, it looks much better than software-based background replacements.

Apple’s technology does a better job of keeping your hair and clothes within focus, but still separated from artificial backgrounds. And best of all, it doesn’t look like a cheap green screen effect. You can choose from a few color gradients, shots of Apple Park or your own pictures or videos.

Other highlights of macOS Sequoia

Here are a few other upgrades I appreciated:

• The Passwords app does a decent job of collecting your stored passwords, but it’s clearly just a first attempt. It’s not nearly as smart about plugging in my passwords into browser fields as apps like 1Password and LastPass.

• The Notes app now lets you record voice notes and automatically transcribes them. You can also continue to jot down text during a voice recording, making it a useful way to keep track of interviews and lectures. I’m hoping future updates add features like multi-speaker detection.

• Being able to jot down math equations in Notes is cool, but it’s not something I rely on daily. I’m sure it’ll be very useful to high school and college kids taking advanced math courses, though.

• Messages finally gets rich text formatting and a send later option. Huzzah!

Finally, a macOS update worth getting excited about

You’d be forgiven for completely ignoring the last batch of macOS updates, especially if you haven’t been excited about Stage Manager or, sigh, widgets. But if you’re a Mac and iPhone owner, Sequoia is worth an immediate upgrade. Being able to mirror your iPhone and its notifications is genuinely useful, and it’s stuffed with other helpful features. And of course, if you want to get some Apple Intelligence action next month, you’ll have no choice but to upgrade. (We’ll have further impressions on all of Apple’s AI features as they launch.)

Sure, it’s a bit ironic that Apple’s aging desktop OS is getting a shot of life via its mobile platform, but honestly, the best recent Mac features have been directly lifted from iOS and iPadOS. It’s clear that Apple is prioritizing the devices that get updated far more frequently than laptops and desktops. I can’t blame the company for being realistic – for now, I’m just glad it’s thoughtfully trying to make its devices play nice together. (And seriously, just bring iPhone mirroring to the Vision Pro already.)

James McGinniss has been obsessed with decarbonization and the energy grid since he was a high schooler over a decade ago. Now, his startup David Energy has a lofty goal: getting the energy grid to run entirely on clean energy in the next 10 years.

Brooklyn-based David Energy is a software-enabled retail energy provider. It sells electricity to small businesses and residential customers in certain states like New Jersey and Massachusetts, not dissimilar to a Southern Company or Pacific Gas & Electric. But unlike incumbents, it’s focused entirely on clean energy. It also helps customers optimize their energy usage and get monetary rewards for saving energy.

The startup has two business streams right now. Its first and larger focus is its small business strategy, which launched in 2022. This approach supplies these companies with clean energy, and also gives companies a dashboard to track their energy use and audit their energy bills to avoid overconsumption.

McGinniss, the co-founder and CEO of David Energy, told TechCrunch the company just launched its residential-focused business in 2023, which is currently available to consumers in Texas who own an electric vehicle or have smart home systems like Nest. David Energy taps into smart home devices and electric car chargers to help them optimize power and notify users when to charge their car.

“Software is uniquely suited to solve these problems automatically for them,” McGinniss said about both commercial and residential customers. “If we can identify a way to get cheaper power, they buy that through us. What they are coming to us for is basically becoming their energy manager through their software platform.”

McGinniss launched David Energy in 2019. He originally wanted to join a company looking to decarbonize the grid but couldn’t find anyone tackling the problem in the way he thought they could. Now, David Energy now serves thousands of customers and has raised more money to scale.

The startup just raised a $23 million Series A-1 round led by Cathay Innovation with participation from existing investors including USV, Keyframe Capital, Equal Ventures, and BoxGroup. McGinniss said the company plans to use the funds to continue expansion into more geographies and to prove repeatable product market fit before raising its next round.

“Anything related to the power grid, each round the broader investment community has gotten smarter and smarter about it,” McGinniss said. “Back in 2019, no one was really aware of the opportunity. By the next time we were raising, people were talking about climate tech and starting to understand that in the electricity sector there is this massive opportunity.”

Simon Wu, a partner at Cathay Innovation, told TechCrunch that they got interested in David Energy because it aligns perfectly with the French venture firm’s climate thesis. The company has big ambitions but also a product line that is out and making money now.

“David Energy was one of the first [companies] saying, ‘Hey let’s try not to make new energy assets, let’s see what we can do or what we have right now,’” Wu said. “If we are able to optimize what we have that hasn’t moved digitally, we can structurally lower costs on energy in a more software-oriented manner.”

David Energy may face some strong headwinds when it comes to scaling. Each state approaches its energy grids differently, and switching to clean energy or managing energy consumption details aren’t necessarily top-of-mind for residents or commercial businesses.

Residential customers also have to end their contracts with a legacy energy provider to join David Energy, which means the company needs to maintain a rock-solid foundation of trust. On the commercial side, McGinniss said that many of David Energy’s customers thus far are from outbound outreach, as there is an education element involved to get people to sign up. He added that many businesses aren’t thinking about switching to clean energy nor realize that tracking their energy consumption can help them reduce it and save money.

Other startups are looking to tackle the grid’s problems, too. Octopus Energy is one that provides clean energy to residents in the U.K. and Texas; it’s raised more than $2.9 billion in venture funding. Arcadia is a late-stage startup looking to decarbonize the electric grid, and it has raised more than $575 million from VCs.

McGinniss acknowledged that David Energy’s strategy is just one approach to fixing the energy crisis. He added that the company is starting to see a flywheel effect from its existing customers leading to new users, and he’s optimistic about the company’s incentives and software approach, too.

“We want to build the new incumbent on the power grid that will be able to handle this new environment that incumbents aren’t equipped to adapt to,” he said.

Nintendo has submitted a mysterious new wireless device to the Federal Communications Commission (FCC) this weekend, and it isn’t the Switch 2 console we’re expecting to be revealed before next April. The CLO-001 model number doesn’t reveal what it actually is, but it appears to be an entirely new product line given the “001” codes used on devices like the original Switch (HAC-001) and DS (NTR-001) consoles.

It’s tagged only as a “wireless device,” not a “wireless game console” or any kind of controller like a Joy-Con. A basic diagram within the documents shows an outline of where the FCC label will be “displayed in a depression area on the bottom” of something with a squarish footprint and rounded-off corners.

The documents also show that there’s no body-worn accessory involved with the testing and that it doesn’t have a battery and can only operate plugged in — it was connected to the same USB-C charger that Nintendo ships with its Switch consoles in tests, but that only reveals it can be powered by USB-C.

The CLO-001 is surprisingly light on wireless technology — there’s no 5GHz or 6GHz Wi-Fi radio and no Bluetooth, but it does sport a 2.4GHz Wi-Fi radio and a 24GHz mmWave sensor. 

The 24GHz mmWave sensor is the most interesting detail we have. It could be a radar sensor to track movement, like closely following gestures or detecting when someone is nearby the device, causing it to turn on automatically like we’ve seen in the 60GHz sensors found in devices like the Pixel 4 and Aqara’s smart home presence sensor.

The Wi-Fi hardware in the first-gen Switch was created by Cypress Semiconductor, a company that was acquired a few years ago by Infineon, which is now promoting the capabilities of its 24GHz radar sensors. 24GHz radar can provide a detection range of up to 100m, with penetration possible through obstacles, at the tradeoff of a larger size and lower resolution, according to Infineon. We couldn’t tell which hardware is inside this device, but the Infineon website includes comparisons to 60GHz sensors.

It really is anyone’s guess right now given the limited information available. But it’s rumor season with the Switch 2 coming, so let us know if you have any better ideas!

A transcript from a key meeting at the firm behind the ill-fated Titan submersible has revealed the CEO said in 2018: “No-one is dying under my watch – period.”

It captures a heated exchange between OceanGate chief Stockton Rush and his former director of marine operations, David Lochridge, plus three other staff.

The log shows Mr Lochridge raised safety concerns, to which Rush responded: “I have no desire to die… I think this is one of the safest things I will ever do.”

The document was released by the US Coast Guard as part of its inquiry into the June 2023 disaster when the sub imploded while journeying to the Titanic shipwreck. All five passengers were killed, including Rush.

OceanGate suspended all exploration and commercial operations following the incident, which led to questions about the submersible’s safety and design.

During two weeks of hearings, investigators are seeking to uncover what led to the tragedy, and to make recommendations to avoid repeat incidents.

The transcript was uploaded to the inquiry website on Friday, but sections of the document were redacted.

The US Coast Guard has now confirmed to BBC News who was speaking in this key exchange during the two-hour meeting.

Mr Lochridge – who gave evidence at the public inquiry last week as a former OceanGate employee – was called to the meeting on 19 January 2018.

He had compiled a “quality inspection report”, which raised serious problems with the sub’s design.

These included concerns about the poor quality of the sub’s hull, which was made of carbon fibre, and issues with the way Titan was being constructed and tested.

He told the inquiry last week: “That meeting turned out to be a two-hour, 10-minute discussion… on my termination and how my disagreements with the organisation, with regards to safety, didn’t matter.”

The 2018 meeting was recorded, and the transcript captures Mr Lochridge saying: “I am addressing what I view as safety concerns, concerns I have mentioned verbally… which have been dismissed by everybody.”

Stockton Rush was recorded replying: “I’ve listened to them, and I have given you my response to them, and you think my response is inadequate.”

Rush went on to say: “Everything I’ve done on this project is people telling me it won’t work – you can’t do that.”

After telling the meeting attendees that he had no desire to die and that he believed his sub was safe, Rush continued by saying: “I’ve got a nice granddaughter. I am going to be around. I understand this kind of risk, and I’m going into it with eyes open and I think this is one of the safest things I will ever do.”

He then added: “I can come up with 50 reasons why we have to call it off and we fail as a company. I’m not dying. No one is dying under my watch – period.”

Mr Lochridge was fired after the meeting and then took his concerns to the US Occupational Safety and Health Administration (Osha).

But he said the US government agency was slow and failed to act. After increasing pressure from OceanGate’s lawyers, he dropped the case and signed a non-disclosure agreement.

At the end of his evidence to the Titan inquiry last week, he said that if the authorities had properly investigated OceanGate, the tragedy would have been averted.

In other developments related to the case, the US Coast Guard has also released an image of Titan showing how its dome fell off as the submersible was lifted out of the sea following a dive in 2021.

A paying passenger who was on that particular Titan mission described the incident during his own testimony on Friday.

Fred Hagen said: “The force of the platform hitting the deck… it basically sheared off several bolts and they shot off like bullets. And the titanium dome fell off.”

This was one of 118 technical incidents listed by the US Coast Guard with Titan dives to the Titanic that took place before the 2023 disaster.

The public hearings continue this week.

Monday’s evidence comes from OceanGate co-founder Guillermo Sohnlein, the company’s former engineering director Phil Brooks and Roy Thomas from the American Bureau of Shipping.

Computer manufacturing giant Dell is looking into claims that its infrastructure was breached and sensitive data on thousands of employees stolen.

Late last week, a threat actor with the alias ‘grep’ posted a new thread on the infamous dark web forum BreachForums. In it, they offered a large Dell database for sale, allegedly containing sensitive employee information.

The algorithms that underpin artificial intelligence systems like ChatGPT can’t learn as they go along, forcing tech companies to spend billions of dollars to train new models from scratch. While this has been a concern in the industry for some time, a new study suggests there is an inherent problem with the way models are designed – but there may be a way to solve it.

Most AIs today are so-called neural networks inspired by how brains work, with processing units known as artificial neurons. They typically go through distinct phases in their development. First, the AI is trained, which sees its artificial neurons fine-tuned by an algorithm to better reflect a given dataset. Then, the AI can be used to respond to new data, such as text inputs like those put into ChatGPT. However, once the model’s neurons have been set in the training phase, they can’t update and learn from new data.

This means that most large AI models must be retrained if new data becomes available, which can be prohibitively expensive, especially when those new datasets consist of large portions of the entire internet.

Researchers have wondered whether these models can incorporate new knowledge after the initial training, which would reduce costs, but it has been unclear whether they are capable of it.

Now, Shibhansh Dohare at the University of Alberta in Canada and his colleagues have tested whether the most common AI models can be adapted to continually learn. The team found that they quickly lose the ability to learn anything new, with vast numbers of artificial neurons getting stuck on a value of zero after they are exposed to new data.

“If you think of it like your brain, then it’ll be like 90 per cent of the neurons are dead,” says Dohare. “There’s just not enough left for you to learn.”

Dohare and his team first trained AI systems from the ImageNet database, which consists of 14 million labelled images of simple objects like houses or cats. But rather than train the AI once and then test it by trying to distinguish between two images multiple times, as is standard, they retrained the model after each pair of images.

They tested a range of different learning algorithms in this way and found that after a couple of thousand retraining cycles, the networks appeared unable to learn and performed poorly, with many neurons appearing “dead”, or with a value of zero.

The team also trained AIs to simulate an ant learning to walk through reinforcement learning, a common method where an AI is taught what success looks like and figures out the rules using trial and error. When they tried to adapt this technique to enable continual learning by retraining the algorithm after walking on different surfaces, they found that it also leads to a significant inability to learn.

This problem seems inherent to the way these systems learn, says Dohare, but there is a possible way around it. The researchers developed an algorithm that randomly turns some neurons on after each training round, and it appeared to reduce the poor performance. “If a [neuron] has died, then we just revive it,” says Dohare. “Now it’s able to learn again.”

The algorithm looks promising, but it will need to be tested for much larger systems before we can be sure that it will help, says Mark van der Wilk at the University of Oxford.

“A solution to continual learning is literally a billion dollar question,” he says. “A real, comprehensive solution that would allow you to continuously update a model would reduce the cost of training these models significantly.”