Posted on

Fresh Nvidia RTX 5090 and 5080 stock rumors again suggest that next-gen GPUs will not be easy to buy

  • Nvidia’s Blackwell stock levels are again rumored to be shaky
  • Launch stock of the RTX 5090 might be very thin on the ground
  • The RTX 5080 should be better, but still might be tricky to find in the early days

Nvidia’s RTX 5000 graphics cards could be in short supply when these next-gen GPUs first arrive, if a fresh rumor is right – and this isn’t the first time we’ve heard chatter from the grapevine along these lines.

This time it comes from a regular source of rumors on YouTube, Moore’s Law is Dead (MLID), who discussed the subject of Blackwell stock in his latest video.

The top-level summary is that the RTX 5090 is looking very shaky indeed for its amount of launch stock, and the news doesn’t sound all that much better for the RTX 5080.

MLID’s first source – season all of this appropriately – works at a US distributor and said that their organization will only have around 20 boards for the RTX 5080, and nothing at all for the flagship. Yes, zero stock for the RTX 5090, which sounds worrying indeed.

The theoretical allocation of 20 or so for the RTX 5080 graphics cards is for the first month of sales (meaning February), and to compare with the last generation, this distributor had a couple of hundred RTX 4080s back at launch in 2022. So we could potentially be looking at a tenth of that stock for the RTX 5080.

The second source is a graphics card maker (presumably in the US) who said that their firm has the same amount of RTX 5090 boards as with RTX 3090 – and if you recall, RTX 3090 stock was vanishingly thin on the ground. As for the RTX 5080, apparently supply is a ‘fraction’ of that seen with the RTX 4080, although this source doesn’t estimate it’ll be quite as bad as a tenth – more like a third to half of that seen with the RTX 4080.

Another source, also a graphics card maker (in the EU), said that the RTX 5090 is looking like it’ll be ‘very rare’ but that the RTX 5080 seemingly has ‘okay’ stock levels, for the graphics card’s initial launch anyway.

{ window.reliablePageLoad.then(() => { var componentContainer = document.querySelector(“#slice-container-newsletterForm-articleInbodyContent-ocsxerdceydCF4TJCq2Dh3”); if (componentContainer) { var data = {“layout”:”inbodyContent”,”header”:”Get daily insight, inspiration and deals in your inbox”,”tagline”:”Sign up for breaking news, reviews, opinion, top tech deals, and more.”,”formFooterText”:”By submitting your information you agree to the Terms & Conditions and Privacy Policy and are aged 16 or over.”,”successMessage”:{“body”:”Thank you for signing up. You will receive a confirmation email shortly.”},”failureMessage”:”There was a problem. Please refresh the page and try again.”,”method”:”POST”,”inputs”:[{“type”:”hidden”,”name”:”NAME”},{“type”:”email”,”name”:”MAIL”,”placeholder”:”Your Email Address”,”required”:true},{“type”:”hidden”,”name”:”NEWSLETTER_CODE”,”value”:”XTR-D”},{“type”:”hidden”,”name”:”LANG”,”value”:”EN”},{“type”:”hidden”,”name”:”SOURCE”,”value”:”60″},{“type”:”hidden”,”name”:”COUNTRY”},{“type”:”checkbox”,”name”:”CONTACT_OTHER_BRANDS”,”label”:{“text”:”Contact me with news and offers from other Future brands”}},{“type”:”checkbox”,”name”:”CONTACT_PARTNERS”,”label”:{“text”:”Receive email from us on behalf of our trusted partners or sponsors”}},{“type”:”submit”,”value”:”Sign me up”,”required”:true}],”endpoint”:”https://newsletter-subscribe.futureplc.com/v2/submission/submit”,”analytics”:[{“analyticsType”:”widgetViewed”}],”ariaLabels”:{}}; var triggerHydrate = function() { window.sliceComponents.newsletterForm.hydrate(data, componentContainer); } if (window.lazyObserveElement) { window.lazyObserveElement(componentContainer, triggerHydrate); } else { triggerHydrate(); } } }).catch(err => console.error(‘%c FTE ‘,’background: #9306F9; color: #ffffff’,’Hydration Script has failed for newsletterForm-articleInbodyContent-ocsxerdceydCF4TJCq2Dh3 Slice’, err)); }).catch(err => console.error(‘%c FTE ‘,’background: #9306F9; color: #ffffff’,’Externals script failed to load’, err)); ]]>

Sign up for breaking news, reviews, opinion, top tech deals, and more.

Remember that these are all rumors around third-party Blackwell graphics cards, so they don’t apply to Nvidia’s own Founders Edition RTX 5090 and 5080 cards.

MLID did hear from a contact at Nvidia, although that person made it clear that they weren’t involved in any discussions related to supply – but did observe that Team Green has warned staff that there won’t be many RTX 5090 Founders Editions available from the employee store at launch. They noted that with the RTX 4090, it was very easy to get one of those (heavily discounted) GPUs from that in-house store.

Nvidia RTX 5090 & 5080 Supply Leak | RX 9070 XT Benchmarks | AMD FSR 4 Support for RDNA 3 – YouTube Nvidia RTX 5090 & 5080 Supply Leak | RX 9070 XT Benchmarks | AMD FSR 4 Support for RDNA 3 - YouTube Watch On

Analysis: Caveats and more optimistic glimmers

Obviously, all this comes with weighty caveats. It’s still only a few sources, even if multiple insiders have chimed in here – and it’s only the one US distributor (others could be faring better, perhaps, especially for the RTX 5080).

Indeed, there’s some mixed chatter here for the RTX 5080, and some indications of stronger stock levels, like that mention of inventory being a third to a half of that seen with the RTX 4080. That doesn’t sound as gloomy as some of the other estimations here, but as MLID points out, RTX 4080 supply wasn’t great, though, and part of the reason it hung around was because this GPU wasn’t very popular. The RTX 5080 could prove a great deal more in demand, and so could still sell out in a relative flash.

We don’t know how much trust to put in this speculation, of course, and the Founders Edition could be different too – there’s no solid evidence on those models. But it certainly makes sense that Nvidia wouldn’t particularly want to prioritize RTX 5090 stock in particular – for its own boards, or third-party graphics cards. Firstly, because AMD RDNA 4 isn’t even remotely competing with Nvidia’s new flagship, and secondly, because Team Green will doubtless want to use the top Blackwell chips for AI rather than the 5090, as that’s where all the big profits lie.

All this doesn’t fill me with confidence about the general picture of RTX 5000 stock, it must be said, particularly as a report from last week chimes pretty much with the assertions here, hinting that it could be a battle to get one of Nvidia’s next-gen GPUs at launch.

You might also like…

Source

Posted on

If you still own an RTX 3000 series GPU, you might be in luck

  • Nvidia’s Bryan Catanzaro suggests older RTX 3000 GPUs could potentially get Frame Generation
  • The new Frame Generation model doesn’t need an Optical Flow accelerator
  • Tensor Cores could be the deciding factor for the RTX 3000 series receiving Frame Generation

With Nvidia’s RTX 5000 series launch on the horizon, it’s easy to be tempted into buying the latest and (hopefully) the greatest, but Frame Generation could change that – it’s not just being improved on RTX 4000 series GPUs and Team Green’s latest Blackwell GPUs like the RTX 5090 (Multi Frame Generation), but potentially RTX 3000 GPUs as well.

Hints that older Nvidia GPUs might finally get Frame Generation, a clever feature that uses AI to generate additional frames to boost overall frame rates, comes from Digital Foundry’s interview (which you can view below) with Nvidia’s Applied Deep Learning Research VP, Bryan Catanzaro – as reported by Wccftech, Catanzaro mentioned Nvidia will be looking at ways to get the best out of older hardware. We know that the current model of Frame Generation will receive improvements later this month once the RTX 5080 and 5090 launch, which will utilize less VRAM while providing better performance thanks to Tensor Cores.

Catanzaro made it clear that DLSS 3 Frame Generation was built upon Nvidia’s Optical Flow hardware accelerator (motion detector for objects between frames), with the RTX 4000 series GPUs maintaining a much more improved version compared to RTX 3000 GPUs – the new model of Frame Generation (and Multi Frame Generation exclusive to RTX 5000 series GPUs) won’t require the Optical Flow accelerator, but rather an AI-based solution.

Since the new model will rely on a higher standard of Tensor Cores (which increases AI performance) which both RTX 4000 and RTX 5000 GPUs come with, it isn’t that simple for Team Green to bring Frame Generation to the older GPUs. With Frame Generation supposedly using much less VRAM usage and not needing an Optical Flow accelerator, however, RTX 3000 users could be in luck (despite the weaker Tensor Cores).

Inside DLSS 4 & Nvidia Machine Learning: The Bryan Catanzaro Interview – YouTube Inside DLSS 4 & Nvidia Machine Learning: The Bryan Catanzaro Interview - YouTube Watch On

Again, I must ask, is there a need to buy an RTX 5000 series GPU?

While this is essentially just speculation about future possibilities, there’s a chance that Nvidia RTX 3000 series users could get the full package of DLSS 4 that includes Frame Generation. DLSS 3 has been available to both RTX 3000 and 2000 series users with super-resolution, DLAA, and ray reconstruction at their disposal – but Frame Generation has so far been exclusive to the RTX 4000 series.

With the requirement of an Optical Flow accelerator now gone, the chances of Frame Generation making its way to RTX 3000 GPUs are now much higher. The main hurdle that could stop this is the weaker Tensor Cores as I mentioned, but the fact that we’re at least getting discussions about it with Catanzaro shouldn’t be taken lightly.

For now, it’s best to look forward to what improvements DLSS 4 will bring to older GPUs and stay patient for any future updates. If Frame Generation for RTX 3000 GPUs does happen, it could breathe new life into the older graphics cards, and could mean gamers that can’t afford a new RTX 5000 series GPU will continue to be able to play new PC games for a few more years.

{ window.reliablePageLoad.then(() => { var componentContainer = document.querySelector(“#slice-container-newsletterForm-articleInbodyContent-TxRkY9knMtvr6ST5qG34cV”); if (componentContainer) { var data = {“layout”:”inbodyContent”,”header”:”Get daily insight, inspiration and deals in your inbox”,”tagline”:”Sign up for breaking news, reviews, opinion, top tech deals, and more.”,”formFooterText”:”By submitting your information you agree to the Terms & Conditions and Privacy Policy and are aged 16 or over.”,”successMessage”:{“body”:”Thank you for signing up. You will receive a confirmation email shortly.”},”failureMessage”:”There was a problem. Please refresh the page and try again.”,”method”:”POST”,”inputs”:[{“type”:”hidden”,”name”:”NAME”},{“type”:”email”,”name”:”MAIL”,”placeholder”:”Your Email Address”,”required”:true},{“type”:”hidden”,”name”:”NEWSLETTER_CODE”,”value”:”XTR-D”},{“type”:”hidden”,”name”:”LANG”,”value”:”EN”},{“type”:”hidden”,”name”:”SOURCE”,”value”:”60″},{“type”:”hidden”,”name”:”COUNTRY”},{“type”:”checkbox”,”name”:”CONTACT_OTHER_BRANDS”,”label”:{“text”:”Contact me with news and offers from other Future brands”}},{“type”:”checkbox”,”name”:”CONTACT_PARTNERS”,”label”:{“text”:”Receive email from us on behalf of our trusted partners or sponsors”}},{“type”:”submit”,”value”:”Sign me up”,”required”:true}],”endpoint”:”https://newsletter-subscribe.futureplc.com/v2/submission/submit”,”analytics”:[{“analyticsType”:”widgetViewed”}],”ariaLabels”:{}}; var triggerHydrate = function() { window.sliceComponents.newsletterForm.hydrate(data, componentContainer); } if (window.lazyObserveElement) { window.lazyObserveElement(componentContainer, triggerHydrate); } else { triggerHydrate(); } } }).catch(err => console.error(‘%c FTE ‘,’background: #9306F9; color: #ffffff’,’Hydration Script has failed for newsletterForm-articleInbodyContent-TxRkY9knMtvr6ST5qG34cV Slice’, err)); }).catch(err => console.error(‘%c FTE ‘,’background: #9306F9; color: #ffffff’,’Externals script failed to load’, err)); ]]>

Sign up for breaking news, reviews, opinion, top tech deals, and more.

You may also like…

Source

Posted on

Where IT comes from: Behind the scenes at Pure Storage’s European R&D centre

You’re a $2.8bn storage supplier with flash arrays at the core of your business. How do you do research and development (R&D), test new products, test customer workload issues, and test array products over years-long timescales for issues that only arise as software, network and application changes concatenate and interact?

Meanwhile, you are a global business with R&D and developer teams across time zones, all at work on ongoing monthly and quarterly updates, and incessant efforts to optimise storage controller software.

The base product is essentially the same, so effective collaboration and information sharing between teams spread across continents is key. But at the same time, you must also test regional customer-specific array configurations.

The Pure Storage solution is to divide responsibility between hardware and software, while also sharing specific R&D and testing capability between three sites.

There is Santa Clara in California, which is its global headquarters and handles hardware and software R&D and testing. There is also Bangalore in India, which only carries out software R&D and testing.

And there is Prague in Czechia, which recently opened its doors to the IT press. Here, we take a look at what goes on behind the scenes in R&D and product testing at Pure Storage (and its nearby array assembly operation).

Capabilities across the three centres are in many ways duplicated, which sounds counter-productive. But it’s not quite as simple as that, according to engineering vice-president and Pure’s Prague site leader Paul Melmon.

“Generally speaking, the same capabilities exist across all sites, except Santa Clara with its hardware development facilities,” he says.

“We try to make projects run autonomously and to minimise cross-time zone meetings,” says Melmon, adding that information can be shared globally in other ways, such as in Git repositories. 

“Lots of companies split projects into many pieces and distribute them,” says chief technology officer Rob Lee. “We choose significant parts of individual products and give them to individual sites, and have product managers for specific products sitting locally.”

“We give a lot of thought to what to centralise and what to not,” says Melmon. “We have rules of engagement that aim at communications that can minimise the number of meetings.”

R&D, testing and talent in Prague

As mentioned, Pure’s Prague site is dedicated to software research, development and testing. It runs thousands of ongoing and custom test routines on hundreds of racks of software. These are divided into “persistent” and “non-persistent” testing, says engineering manager Tom Healy.

Non-persistent is ephemeral. It tests for issues in specific customer deployment configurations, or the impact of updates on controller code.

Persistent testing is long-term. It can be very long-term, in fact, with racks in place with, for example, generation upon generation of Pure Storage FlashBlade file and object storage deployed.

“Sometimes things can take years to occur,” says Healy.

“Our testbeds include, for example, FlashBlade capacity that dates back to the first generation [2016], some virtualisation, and a Windows application platform. All of this will run for years, to follow the lifecycle of customer systems and to test for the effect of changes to software and hardware, and its stability,” says Healy.

“And, FlashBlade uses Ethernet, so we are checking what happens when changes happen in the workload, simulation of new cabling, media, hitting it with broadcast storms, simulating signal degradation, etc,” he adds. 

Meanwhile, at the Prague facility, hundreds of engineers work constantly on storage array software to meet ongoing monthly and quarterly updates. 

Pure’s Prague R&D facility has just celebrated its five-year anniversary. It is resident in the Amazon building (no relation) and others in the riverside Karlin district. There it employs 600 people – 50% Czech, 50% from elsewhere – with up to 50 nationalities on-site.

Prague was chosen as a European centre because of its proximity to so many of Pure’s customers, but also, says Melmon, because of the availability of talent. “It’s on a level with Silicon Valley,” he says, and its accessibility in terms of transport links, universities, graduates in computer science, cost of living and general likeability of the city.

Speaking of the River City complex in which Pure is located, Melmon describes it as having a “South of Market” feel, referring to the fashionable area of San Francisco that became a honeypot for startups in the 1990s. “It’s the place to be if you’re in tech and AI [artificial intelligence]. There are meetups in the evening. It’s the cool new place to be in Prague.”

But it’s not just a cool place to work and live. Melmon points to the 19.7% figure, which is the proportion of revenue Pure spends on R&D. 

Prague is the biggest Pure Storage R&D centre outside the US and has delivered about a third of its FlashArray product development. Meanwhile, FlashBlade//S was jointly designed and tested there, while key elements of the Pure Fusion workload management platform and its Pure1 AIOps were developed in the Czech capital. Meanwhile, 100% of Portworx Data Services and Pure’s disaster recovery as a service (DRaaS) offering came from there too.

That’s the result, with Prague as a key pillar in the three-site R&D and testing strategy of Pure Storage.

Nearby, also in Czechia, is one of its global assembly centres, which you can read about here.

Source

Posted on

Open-R1 is a truly open version of DeepSeek AI

On Monday, DeepSeek R1 crashed the stock market once it became clear to some of the investors trading AI-related stocks that the Chinese startup had found a way to train AI as capable as ChatGPT o1 without access to the state-of-the-art NVIDIA chips that OpenAI and US AI firms have access to. That’s why firms creating hardware for AI infrastructure suffered the most. NVIDIA shed nearly $600 billion in market cap, while the entire market lost almost $1 trillion.

I said at the time that the reactions might be blown out of proportion. Yes, DeepSeek employed software optimizations to develop AI as capable as o1 instead of relying on hardware. But that doesn’t mean NVIDIA’s GPUs are suddenly obsolete. It just realigns the playing field while providing a new way to innovate.

I still think that AI firms with access to the latest hardware and top-tier software talent will have an edge over Chinese rivals. All a company like OpenAI or Google has to do is replicate some of the tricks DeepSeek used to match the Chinese startup’s AI training and usage efficiency and then leapfrog it. The latest AI chips will still be very important here.

It turns out it’s not just the big AI firms that might try to copy what DeepSeek has done. A team of developers calling themselves Open-R1 wants to replicate the DeepSeek R1 success to create a reasoning AI model that’s just as powerful as R1. There’s a big twist in all of this that AI fans in Western markets will appreciate. Open-R1 should be even more transparent than DeepSeek R1.

Tech. Entertainment. Science. Your inbox.

Sign up for the most interesting tech & entertainment news out there.

By signing up, I agree to the Terms of Use and have reviewed the Privacy Notice.

DeepSeek’s decision to make its AI models open-source was brilliant. This ensured that anyone could access and install the model on their computer. From there, they’d have a local model as capable as ChatGPT o1. The open-source route would also drive up adoption and testing. News about R1’s capabilities would spread rapidly.

But, as the Open-R1 researchers explain on Hugging Face, DeepSeek R1 isn’t fully open-source:

The release of DeepSeek-R1 is an amazing boon for the community, but they didn’t release everything—although the model weights are open, the datasets and code used to train the model are not .

That’s where Open-R1 is coming in: 

The goal of Open-R1 is to build these last missing pieces so that the whole research and industry community can build similar or better models using these recipes and datasets. And by doing this in the open, everybody in the community can contribute!

Specifically, the Open-R1 team wants to answer the following questions about DeepSeek R1 while they develop an identical AI:

Data collection: How were the reasoning-specific datasets curated?

Model training: No training code was released by DeepSeek, so it is unknown which hyperparameters work best and how they differ across different model families and scales.

Scaling laws: What are the compute and data trade-offs in training reasoning models?

The researchers plan to clone DeepSeek’s development strategy for R1, further fine-tune it, and create a truly open-source Open-R1 model that anyone could use.

Interestingly, the Open-R1 researchers want to distill DeepSeek R1 and create a high-quality reasoning dataset. DeepSeek might have done its own distillation, with OpenAI claiming the Chinese startup used ChatGPT to train its earlier versions of AI. That work might have been critical to getting to DeepSeek R1. It’s unclear if OpenAI can prove these allegations with absolute certainty.

However, the Open-R1 researchers have their own strategy after distilling R1, with the blog explaining how they plan to go forward.

If successful, Open-R1 could be a stepping-stone for developing other sophisticated AI models, and anyone could do it. The advantage here is that you would not have to go through the same training process. Conversely, that’s what OpenAI says DeepSeek did with ChatGPT, using some of its outputs to save money on training the AI.

An open-source reasoning model like the Open-R1 model the researchers propose could be used for other purposes, not just math and coding. The researchers mention medicine, where reasoning AI “could have significant impact.”

That said, it’s unclear how long the project will take and when Open-R1 will be ready for testing. Other AI researchers interested in Open-R1 can check out the project on GitHub.

Source

Posted on

iOS 18.3 coming soon: New features, release date, AI updates, more

Apple is about to end the beta cycle of iOS 18.3. Unlike previous software updates, which took a lot more time to release, this one will likely be pretty straightforward. In this article, you’ll discover everything new with this future operating system update, its possible release date, and all the devices compatible with it.

Release Date

With iOS 15.3 and iOS 16.3, Apple only seeded two betas before making a Release Candidate version available. In early 2024, the iOS 17.3 beta needed an extra release, as Apple seeded three betas before the Release Candidate version. Apple then released iOS 17.3 in the third week of January.

With iOS 18.3, Apple seeded three betas and a Release Candidate version so far. We expect this software update to be available as soon as January 27. That said, Cupertino has a much bigger iOS 18.4 update coming in April.

iOS 18.3 beta 1 features

The first beta of iOS 18.3 only includes a few new features. Here’s what we know so far:

Tech. Entertainment. Science. Your inbox.

Sign up for the most interesting tech & entertainment news out there.

By signing up, I agree to the Terms of Use and have reviewed the Privacy Notice.

Robot vacuum support: This feature was delayed to 2025. X user Aaron Perris discovered strings in iOS 18.3 beta code that show Apple wants to offer this feature with this next software update. When it’s available, robot vacuum makers will be able to control their vacuums through Apple’s Home app.

Turn off satellite communication: According to Bloomberg’s Mark Gurman, those companies who deploy iPhones through Mobile Device Management will be able to disable satellite communication services. The journalist says this is “designed for defense/space agencies who don’t want employees tapping into it.”

Camera Control icons: Apple has updated the Camera Control icons to the new Dark Mode. This change has been spotted by X user Aaron Perris.

AI Updates? At the moment, it’s unclear what changes Apple Intelligence might have in iOS 18.3. Given how far behind Apple is when it comes to AI, however, it’s possible that Apple will try to slip some new features and improvements into iOS 18.3.

iOS 18.3 beta 2 features

CarPlay featuresImage source: Porsche

Three weeks after Apple seeded the first beta of iOS 18.3, the company released a new build. At first, it doesn’t look like there are new features available with this version. However, 9to5Mac found under-the-hood changes.

CarPlay 2.0: It seems the latest beta keeps adding references to the new CarPlay experience. Apple missed its 2024 deadline, and since iOS 18.2, we’ve seen Cupertino adding new files about this upcoming feature to its software. With the latest iOS 18.3 beta 2, the company added references to “CarPlayHybridInstrument” in the Maps app.

Calculator bug: iOS 18 removed the ability to repeat math operations. However, the latest beta has restored this feature.

iOS 18.3 beta 3 features

As Apple comes close to the public release of this software, the company added a few more tweaks to this upcoming update:

Notification Summary: After complaints that Apple Intelligence was hallucinating while summarizing news headlines, the company decided to turn off summarization for the News category. In addition, Apple now notes that this might happen.

Messages: The iMessage menu now features a new Genmoji toggle, which helps users access this feature faster.

Camera Control tweak: The recently-added AE/AF Lock has been renamed to “Lock Focus and Exposure.”

iOS 18.3 RC Release Notes

These are the release notes of this update:

Notification summaries (All iPhone 16 models, iPhone 15 Pro, iPhone 15 Pro Max)

  • Easily manage settings for notification summaries from the Lock Screen
  • Updated style for summarized notifications better distinguishes them from other notifications by using italicized text as well as the glyph
  • Notification summaries for News & Entertainment apps are temporarily unavailable, and users who opt-in will see them again when the feature becomes available

This update includes the following enhancements and bug fixes:

  • Calculator repeats the last mathematical operation when you tap the equals sign again
  • Fixes an issue where the keyboard might disappear when initiating a typed Siri request
  • Resolves an issue where audio playback continues until the song ends even after closing Apple Music

Device compatibility

Image source: José Adorno for BGR

iOS 18.3 is compatible with the following devices:

  • iPhone XR, XS, and XS Max
  • iPhone 11
  • iPhone 11 Pro and 11 Pro Max
  • iPhone SE (2nd gen)
  • iPhone 12 mini and iPhone 12
  • iPhone 12 Pro and iPhone 12 Pro Max
  • iPhone 13 mini and iPhone 13
  • iPhone 13 Pro and iPhone 13 Pro Max
  • iPhone SE (3rd gen)
  • iPhone 14 and iPhone 14 Plus
  • iPhone 14 Pro and iPhone 14 Pro Max
  • iPhone 15 and iPhone 15 Plus
  • iPhone 15 Pro and iPhone 15 Pro Max
  • iPhone 16 and iPhone 16 Plus
  • iPhone 16 Pro and iPhone 16 Pro Max

Source

Posted on

Over 40 journalists and lawyers submit evidence to PSNI surveillance inquiry

Over 40 journalists and lawyers have submitted evidence to an independent review into allegations that the Police Service of Northern Ireland unlawfully spied on journalists and members of the legal profession.

The review, commissioned by the Chief Constable of Northern Ireland Jon Boutcher, is inquiring into allegations that the PSNI collected the phone data of lawyers, journalists and NGOs, breaching journalists’ confidential sources and legal privilege between lawyers and clients.

The inquiry follows disclosures by the Investigatory Powers Tribunal that the PSNI and the Metropolitan Police had unlawfully carried out surveillance operations against journalists Barry McCaffrey and Trevor Birney.
 
It comes as MPs in the Northern Ireland Affairs select committee are preparing to take evidence next week from the two journalists as part of a one-day hearing into press freedom in Northern Ireland.

Angus McCullough KC, who is leading the review into allegations of PSNI surveillance over 14 years between January 2011 and December 2024, disclosed in a progress report today that 50 individuals and two organisations have submitted evidence to the review. Some 80% of the responses came from journalists and lawyers. A small number, representing 5%, raised issues that fell outside McCullough’s terms of reference.

There was a presumption, he wrote in a progress report, that individuals who were subject to improper or unlawful surveillance would be informed “subject to legal constraints”.

McCullough, assisted by Matthew Hill, who took part in the Post Office Horizon inquiry, and Rajkiran Arhestey, said he was confident that the review team had been given unrestricted and unsupervised access to PSNI documents and computer systems.

According to the progress report, investigators said they were keen to understand what scope there may be for police officers to bypass the proper statutory processes and whether there was a likelihood of “off the books surveillance”. 

McCullough said that surveillance need not be targeted directly against journalists or lawyers for it to have a tendency to reveal legally privileged or journalistic material, including journalistic sources.

The review team has been able to use information in submitted evidence to identify keywords or specific names and contact details to conduct searches of PSNI computer systems to identify cases of surveillance.

“We have also taken steps to ensure, so far as it is possible, that the PSNI are not able to see what we have been searching for,” the report states, in order to guarantee the anonymity of people who have submitted evidence to the review.

Searches of the PSNI systems were “considerably more time consuming” than predicted. Legislation on surveillance and the PSNI’s systems and processes, and repositories for storing data had changed considerably over 14 years. 

McCullough said that he rejected proposals to expand the review to assess the PSNI’s monitoring of its own police officers and staff, and could not consider surveillance by MI5, the British Military, or other police forces.

The Investigatory Powers Tribunal found in December 2025 that a former chief constable of the PSNI, George Hamilton, acted unlawfully by signing off on a directed surveillance operation to identify the suspected source of the two Northern Ireland journalists.

Birney and McCaffrey were unlawfully arrested and their property seized after they produced a film exposing police collusion with a paramilitary group that killed 6 innocent people in Loughinisland, County Down, in 1994.

Court received 60 complaints on PSNI in 3 years

The Police Service of Northern Ireland has been referred in complaints UK’s surveillance court, the Investigatory Powers Tribunal (IPT) nearly 60 times between 2022 and 2024.

The tribunal has ordered the PSNI to disclose internal documents 28 times, and has made 4 adjudications in cases involving the PSNI over the same period, according to figures disclosed by the Northern Ireland Policing Board.

Chief constable, Jon Boucher, admitted in a report published during the course of the legal proceedings that it had placed over 500 lawyers and 300 journalists under surveillance. Those targeted included more than a dozen journalists working for the BBC.

The IPT is considering a claim that the PSNI unlawfully spied against former BBC journalist, Vincent Kearney, during his work on a 2011 Spotlight documentary investigating the independence of the police watchdog in Northern Ireland.

Campaigners have claimed the McCullough review does not go far enough and have called for the government to set up a public inquiry into police surveillance of journalists in Northern Ireland and the rest of the UK.

Computer Weekly has submitted evidence to the McCullough review.

Source

Posted on

NAO: UK government cyber resilience weak in face of mounting threats

The National Audit Office (NAO) has found the UK government’s cyber resilience to be significantly behind where it needs to be, in the face of mounting and more dangerous threats.

In its Government cyber resilience report, the public spending watchdog warned that the cyber threat to the UK government is “severe and advancing quickly”. It found that 58 critical government IT systems, assessed in 2024, had significant gaps in cyber resilience, and the government does not know how vulnerable at least 228 “legacy” IT systems are to cyber attack.

The report does not cover the cyber resilience of local government, the NHS, or the nation as a whole. Fieldwork for the report was conducted between May and October 2024, with NAO staff interviewing officials from the Cabinet Office about efforts to support government departments in the implementation of the Government Cyber Security Strategy: 2022-2030.

The strategy included a target for key government organisations to be “significantly hardened to cyber attack by 2025”, but the government has not improved its cyber resilience fast enough to meet this aim, said the NAO.

The NAO also interviewed officials from the National Cyber Security Centre (NCSC) and the Central Digital and Data Office (CDDO), along with cyber security civil servants from government departments and the British Library.

The biggest risk to making the UK government resilient to cyber attack is a yawning skills gap, according to the report. It found one in three cyber security roles in government were vacant or filled by temporary – and more expensive – staff in 2023-24, while more than half of cyber roles in several departments were vacant, and 70% of specialist security architects were staff on temporary contracts.

The NAO said departments reported that salaries and civil service recruitment processes are barriers to hiring and keeping people with cyber skills.

Other concerns include a lack of coordination within government, which is jeopardising effective cyber defence. The NAO found that the respective roles of departments and central organisations, such as the NCSC, are “insufficiently understood”, and nor have departmental leaders “consistently recognised the relevance of cyber risk to their strategic goals”.

The government must act now, urged the report’s authors.

Gareth Davies, head of the NAO, said: “The risk of cyber attack is severe, and attacks on key public services are likely to happen regularly, yet government’s work to address this has been slow.

“To avoid serious incidents, build resilience and protect the value for money of its operations, government must catch up with the acute cyber threat it faces.

To avoid serious incidents, build resilience and protect the value for money of its operations, government must catch up with the acute cyber threat it faces Gareth Davies, National Audit Office

“The government will continue to find it difficult to catch up until it successfully addresses the long-standing shortage of cyber skills, strengthens accountability for cyber risk, and better manages the risks posed by legacy IT.”

Cyber resilience gaps

The NAO evaluated whether government is keeping pace with the rapidly evolving cyber threat it faces from hostile actors. It found that it is not.

It spotted that the government’s cyber assurance scheme, GovAssure, which had independently assessed 58 critical departmental IT systems by August 2024, found significant gaps in cyber resilience, with multiple fundamental system controls at low levels of maturity across departments. GovAssure assesses the critical systems of government organisations. It was set up in April 2023.

According to the NAO report, government departments were using at least 228 legacy IT systemsas of March 2024, and the government does not know how vulnerable these systems are to cyber attack.

The report noted that in April 2024, the Cabinet Office Government Security Group (GSG) reported to ministers that some departments had significantly reduced their cyber security improvement programmes to fund other priorities. This was due to “cuts to programme funding, lack of access to cyber skills, challenges with delivery partners, and delays in departmental and cross-government approvals”.

As examples of how damaging cyber attacks can be, the NAO cited the instance, in June 2024, of an attack on a supplier of pathology services to the NHS in south-east London, which led to two NHS foundation trusts postponing 10,152 acute outpatient appointments and 1,710 elective procedures. It also cited the British Library ransomware attack in October 2023, which has already cost £600,000 to rebuild its services. The library expects to spend many times more as it continues to recover.

The report also gave other examples of attacks on the Ministry of Defence and Parliament. In May 2024, the MoD’s payroll contractor’s network was compromised by an attacker – a network that held armed forces staff members’ data. Further back in time, in 2021, a Chinese state-affiliated attacker was, said the report, highly likely responsible for a cyber campaign against the parliamentary email accounts of members across both Houses of Parliament.

The report stated that in March 2024, departments did not have fully funded plans to remediate around half of government’s legacy IT assets – 53%, or 120 out of 228.

The NAO recommends the government develops, shares and starts using a cross-government implementation plan for the Government Cyber Security Strategy within the next six months. It also suggests the whole of government needs to operate differently.

Within the next year, the government should make and enact plans to fill cyber skills gaps in workforces, said the NAO.

Of the technology trumpeted most by the current and previous government – artificial intelligence (AI) – the report said: “AI can improve government’s cyber security, but it can also help threat actors looking to interfere or undermine trust in our democratic system. The NCSC is collaborating with its partners to realise the benefits of AI and protect against the associated security risks.”

Source

Posted on

Sir Alan Bates has ‘serious concerns’ over Post Office scandal compensation budget

Sir Alan Bates has “serious concerns” that the government’s budget for Post Office scandal compensation will be stretched to cover the legal costs of those making claims.

This comes as he welcomed a “very positive and quite insightful” report from the parliamentary business and trade select committee.In its report, the committee called for legally building deadlines for subpostmaster redress with financial penalties for failure.

In its Post Office and Horizon scandal redress: Unfinished business report, the committee also demanded the Post Office be removed from administering any of the redress schemes, up-front legal advice to be offered to claimants, and the appointment of independent adjudicators.

It said lawyers being paid by taxpayers should be instructed to speed up payments to subpostmasters, reduce delays, give the benefit of the doubt to claimants and publish figures on government spending on lawyers.

In February 2024, during a mammoth five-hour business and trade select committee hearing, MPs heard that the complexity and unfairness of schemes for the financial redress of former subpostmasters is leading to slow and often unfair settlements. Witnesses, including Bates, called for a legally binding deadline on when payments should be made.

A year on, and payments are still too slow. In her Autumn Budget statement in October, chancellor Rachel Reeves announced Post Office Horizon scandal compensation funding of £1.8bn.

As of November, just £499m of the £1.8bn has been paid out across four redress schemes, with 72% of the budget for redress still not paid.

Lawyers’ fees

In the report, the committee referred to oral evidence provided by Carl Creswell, director for Post Office policy and business engagement at the department of business and trade, which states that the Budget 2024 allocation of £1.8bn to settle redress costs includes claimants’ lawyers’ fees.

“I have very serious concerns about subpostmaster legal costs being taken out of the financial redress pot,” said Bates. “That money should be ring-fenced for financial redress for victims, not paying their legal costs.”

The business and trade committee said it is “imperative” that claimants are offered legal advice up-front, at no cost to themselves but paid for by the scheme administrators.  

“Years on from the biggest miscarriage of justice in British legal history, thousands of Post Office Horizon victims still don’t have the redress to which they’re entitled for the shatter and ruin of their lives … we can’t go on like this,” said chair of the committee Liam Byrne MP. “Justice delayed is justice denied.

“Victims should have up-front legal advice to help make sure they get what’s fair,” he said. “We need hard deadlines for government lawyers to approve the claims with financial penalties for taking too long. Crucially, we need the Post Office, which caused this scandal in the first place, taken out of the picture.”

Bates said the committee report shows that “at least the politicians involved can see the problems, although the civil servants in the department are blind to them”.

Recommendations welcome

Neil Hudgell at Hudgell Solicitors, which represents hundreds of former subpostmasters seeking compensation, said: “We welcome any recommendations to speed up redress, and many of these seek to remove unnecessary obstacles to justice we have seen over the past few years, repeated across hundreds of cases.

“We have repeatedly said claimants should have access to up-front legal advice in all scheme claims, that they should be given the benefit of the doubt where written evidence is limited given the timeframes we are talking, and that offers should be made at the top of the range for each category of loss. Sadly, we’ve not seen that across most schemes to this point, certainly not with any level of consistency.”

The Post Office scandal was first exposed by Computer Weekly in 2009, revealing the stories of seven subpostmasters and the problems they suffered due to the accounting software (see timeline of Computer Weekly articles about the scandal below).

Source

Posted on

Foodora tests drone and robot deliveries in Sweden

Foodora Norway, the Oslo-based subsidiary of online quick-commerce (Q-commerce) food delivery brand Foodora Group, has partnered with Nordic communications company Tele2 to pilot test home deliveries using robots and drones supported by 5G internet of things (IoT) technology.

The Stockholm metropolitan area was chosen as the test location for the ground robot and drone delivery trials. The Doora robot will be used in the ground tests while the drone trials are being conducted as part of the parallel Foodora Air project.  

A subsidiary of Berlin-based Delivery Hero, Foodora is currently present in Norway, Sweden and Finland, where the company delivers restaurant food, groceries and other consumer goods using cars and bicycles. Foodora also operates in Austria , Hungary and Czechia.

The collaboration with Tele2 forms part of Foodora’s two-year plan to roll out GPS-based robot home deliveries across the Nordic countries by in 2025 and 2026. The ambitious joint venture with Tele2 is focused on self-driving robots and drones connected to 5G and IoT technologies.

Foodora is hoping to roll out commercial Doora ground robot deliveries in Sweden and Norway by year-end 2025, with drone services introduced in 2026.  

Scandinavian countries are lagging behind the US and Finland as regards the development of self-driving robot food delivery services, said Prashant Søegaard, CEO at Foodora Norway. “Our partnership with Tele2 in Sweden will help identify the opportunities, challenges and physical obstacles we face in making Doora work as a viable and popular form of food delivery across Nordic markets,” he added.

In Finland, S-Group and its S-Market supermarket chain partnered with the Tallinn-based Starship Technologies in 2023 to roll out the Starship Robot. The first autonomous S-Market delivery robots (ADRs) became operational in selected towns during December 2023. Monitored in real-time for disruptions to service, the electric-powered Starship ADRs use artificial intelligence (AI) and GPS to plan routes and complete delivery missions.

Mobile app

The Starship Robot platform, over the five-month operating period to 30 May 2024, delivered 150,000 orders to households. Working off S-Market’s dedicated home delivery mobile app, the service had linked to an initial 100 S-Market grocery stores by the end of November.

The platform currently serves 14 towns and cities across Finland , including Helsinki , Tampere and Espoo . S-Group is estimating full-year 2024 deliveries exceeding 400,000 homes.

Additionally, S-Market is recording downloads of over 20,000 per month for the home delivery app, said Tiina Meyer, a senior business developer of retail ecommerce at S-Group.  

“The robot delivery service is a major retailing innovation boosting our grocery business,” she said. “Because of its convenience, it’s hugely popular. Customers appreciate the convenience of this new technology, the speed at which their orders are processed and delivered, the aesthetically pleasing robots, and especially enjoy the music the robots play upon delivery.”

Starship Technologies was incorporated in 2014 by Janus Friis and Ahti Heinla, the two Estonian co-founders of Skype. Headquartered in San Francisco , the company operates engineering subsidiaries in Tallinn and Helsinki .

Foodora’s pilot trials of the ground delivery Doora in Sweden include specific tests related to how the robot navigates pedestrian crossings and curbs, as well as how the robot recognises its surroundings and communicates.

Test and evaluation

The Doora and Foodora Air projects will help to test and evaluate the full potential of the delivery systems when connected to Tele2’s 5G network and IoT technology, said Stefan Trampus, the executive vice-president of Tele2 B2B.

“Within this project, we will be able to use the robot delivery concept to look at other industries and areas of use,” he said. “We can track how GPS and IoT function, and look to optimise solutions. The 5G IoT technology is essential for the drones’ smooth operation. The target is to achieve rapid response times and substantial data transfer capabilities to ensure safe delivery management.”

The Doora ground robot is designed to transport food and other products up to a maximum weight of 20kg. The delivery robot has a top speed of 6km (3.72 miles) per hour and a battery life of eight hours, with a four-hour charge.

The Foodora Air project’s core focus is on testing the operability of a fleet of battery-powered drones, utilising 5G technology provided by Tele2, to deliver meals from various restaurants located in the Stockholm suburb of Värmdö. The Nimbi drones, developed by Swedish firm Aerit, are integrated into Foodora Air’s technology platform.

The all-weather certified Nimbi drones used by Foodora Air are able to operate within a maximum delivery range of 21km and emit 2g of carbon dioxide per kilometre. The Nimbi has a maximum payload carrying capacity of 4kg (10lb), and features a proprietary winching system that enables package pick-up and drop-off without the need for supporting infrastructure.

As part of the Foodora Air trial, deliveries are being restricted to customers’ properties, including front and back gardens where packages can be safely lowered by cable from airborne drones.

The results, and shared test data, emanating from the Doora robot and Foodora Air trials in Sweden will help to shape how the company rolls out drone and robot-supported delivery services in Norway , said Søegaard. “We are closely following the tests in Sweden and hope to learn from that experience,” he said. “The data we collect from the project and trials will better inform us about the challenges we face in Norway and need to overcome and solve.”

Foodora has opened a dialogue with public and private players to expand robot delivery services across Norway . The first significant hurdle will be to obtain authorisation from the Norwegian Public Roads Administration (Statens Vegvesen), and local municipalities, ahead of launching autonomous robots, and at a later stage, drones, in Norway.

Foodora is hoping to roll out a robot delivery service in Norway during the first half of 2025, in collaboration with Danish autonomous air and ground vehicle firm Holo.

In August, Holo partnered with Foodora Norway to conduct a demonstration test at Fornebu, a suburb of Oslo , using a six-wheeled delivery robot supplied by California-based Cartken. The Cartken robot is equipped with three 5G modems and connectivity to multiple mobile operators to ensure continuous communication to shield against signal drop-off. The Cartken uses a PIN code system enabling customers to unlock and retrieve food and other deliveries.

Holo also collaborated with Posten , Norway ’s state-controlled postal service, on a pilot trial to test electric-powered delivery robot Ottobot. The robot was supplied by New York-based Ottonomy.IO. The test, which was conducted in the Oslo district of Filipstadkaia in November and December 2022, has so far not resulted in a decision by Postern to procure delivery robot units for commercial use.  

The Ottobot pilot test was run as a partnership project with AMOI, Norway ’s largest digital marketplace that connects specialist retailers, including restaurants, with customers ordering home delivery.

Posten has been something of a trailblazer in Norway as regards the testing of ground robot delivery systems. In 2018, the company conducted a pilot trial in Oslo on the Buddy Mobility autonomous parcel delivery robot. However, high operating costs and low demand for the service resulted in Posten shuttering the project in 2019.

Source

Posted on

Apple’s hallucinated AI News summaries were just disabled in iOS 18.3 beta 3

Hallucinations are a part of the early genAI experience. Since the early days of ChatGPT, we have warned that AI will make mistakes and that you should always look for sources and check whether its claims are accurate. As hard as they might have tried, the big tech players were not spared.

Google’s AI Overviews in Search delivered advice on how to put glue on pizza and hallucinate other information, forcing Google to deal with the PR mess that followed and fix the AI before releasing it to a wider audience.

Apple wasn’t spared the hallucination humiliation either, with Apple Intelligence conflating News reports to deliver fake information via the summarization feature for the News app. Apple has decided to pull the feature from the latest iOS 18 beta and deploy the needed fixes.

“With the latest beta software releases of iOS 18.3, iPadOS 18.3, and macOS Sequoia 15.3, Notification summaries for the News & Entertainment category will be temporarily unavailable,” an Apple representative told CNBC.

Tech. Entertainment. Science. Your inbox.

Sign up for the most interesting tech & entertainment news out there.

By signing up, I agree to the Terms of Use and have reviewed the Privacy Notice.

Apple is working on improving the News summarization feature, which will return in a future software update. It’s unclear when the feature will be back, but Apple Intelligence continues to be a top priority for Apple’s software development teams.

As a reminder, iOS 18.4 will bring another set of AI features to iPhone, iPad, and Mac, including the smarter Siri that can control some apps and access more user data on the device to provide more helpful assistance.

Apple Intelligence’s hallucination problems went viral in December when the AI summarized several BBC reports into a single notification that started with “Luigi Mangione shoots himself.” Mangione is the alleged Brian Thompson assassin who did not shoot himself.

Other hallucinations date back to November when the AI might have shown some users The New York Times summaries that claimed Israeli Prime Minister Benjamin Netanyahu had been arrested. That wasn’t the case.

Hallucinating news is a big problem for any AI product, whether Apple Intelligence or ChatGPT. After all, we’ve been worried about AI misleading users with fake information and the manipulation risks that might come from AI products controlled by nefarious actors. Companies like Apple must get any news-related AI features right, especially summarization. Either that or not do it at all.

In addition to disabling Notification Summaries for the News and Entertainment category in iOS 18.3 beta 3, Apple has added a label to the feature noting that summaries can contain errors, as the app is in beta.

Apple made another change to how summaries appear in notifications so you can tell them apart from regular notifications. Starting with iOS 18.3 beta 3, they’ll be italicized.

Finally, Apple Intelligence users who install the latest beta can decide whether to enable or disable summaries for an iPhone app directly from the Lock Screen. Swipe to the left on a Notification Summary to get an options menu that will let you disable them for specific apps. The alternative is going into the Settings app, where you’d have customized the AI summarizations before the new beta.

As an iPhone user in the EU, I still can’t get Apple Intelligence. I couldn’t test it or experience any of the hallucination issues that US iOS 18 beta testers have encountered. Hopefully, the hallucination problem will go away by the time I get my hands on Apple Intelligence on the iPhone.

However, it’s great to see Apple admit the errors and pull the AI summarization feature entirely rather than proceeding with it without a proper fix.

Source