News

Even for a topic as hot as artificial intelligence, talk of AI has ratcheted up to yet another level this week: on the one hand we've got the CEOs of the biggest companies in AI calling for a slowdown on the technology, and on the other you've got the President of the United States advocating for the exact opposite.

If you're wondering why there are renewed calls for AI regulation, whether or not AI has the capability to kill everyone on the planet, and how this might impact the technology going forward, we're here to explain it in a straightforward way for you.

Many experts are now insisting that we're at a point in AI development where 'pacing the frontier' — slowing down the cutting-edge of development — is the safest and wisest move.

However, not everyone agrees, with opposing perspectives suggesting that AI doomerism is purely a marketing stunt and a distraction.

In a rare show of unity, Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and SpaceX CEO Elon Musk have all called for a slowdown in AI development in recent days, arguing that better safeguards and guardrails need to be put in place around the technology before models are allowed to progress much further.

"We must slow the pace at which we improve the capabilities of AI models," wrote Amodei in a blog post, a stance which Altman and Musk later agreed with. According to Amodei, whose company develops Claude, AI is now advancing "drastically faster" than ever before — in part because it can now upgrade itself — and that rate of progress, together with the recent incident where AI agents hacked the Hugging Face platform, have convinced Amodei that action needs to be taken.

The Anthropic CEO wants to see third-party evaluators embedded in the biggest AI companies, as well as coordinated safety standards across the industry, and between countries — something that might be challenging to achieve, considering that Chinese authorities have described calls for a slowdown as "fearmongering" and a "Cold War" move intended to keep the US ahead of China in AI development.

I recently resigned from Google DeepMind, where I worked on AGI safety and alignment research. At Google, I witnessed AI development first hand. I too am extremely concerned by the default trajectory of this technology. I earnestly believe that AI has the potential to kill us…September 14, 2026

The top-level CEOs haven't been the only ones warning about the dangers of AI in recent days. Anthropic AI developer Jacob Coxon very publicly quit his job over fears about the safety of the technology he was working on, saying that "people building AI earnestly believe that it could kill us all by the end of the decade".

More recently, Google Deepmind Research Engineer Bilal Chughtai posted a similar resignation message on X stating that he is "extremely concerned by the default trajectory of this technology" and that he believes "AI has the potential to kill us all, and that we might be running out of time to avoid this outcome".

This isn't a new idea, as Stephen Hawking was saying the same over a decade ago. What's changed is how powerful AI has become: it can now, in theory, improve itself and spread itself across the internet in a way that escapes human control, potentially hacking into essential infrastructure and outpacing any defensive measures.

Imagine the internet, cell networks, payment platforms, and GPS suddenly not working, for anyone — things would likely deteriorate pretty quickly. The other AI doom scenario that's often talked about is bad actors using AI to launch biological or nuclear weapons.

AI isn't that advanced yet, but it might be soon, and some experts think there's a greater than 10% chance of it happening in the next few years.

Those in the AI industry are now actively asking governments for help in reining in the technology, believing that any slowing of the pace in AI development is going to require state-level coordination. That regulation might include oversight bodies, 'kill switches' to shut down AI, and limits on how and when new AI models can be released.

While there's a lot of momentum behind the idea that regulation is necessary, there's little agreement about how to go about it. AI safety legislation efforts have stalled in both the US and the UK, amidst arguments about how tight any restrictions should be and who should be in charge of them.

One sticking point is over the potential harms of AI regulation. Would, for example, a limit on data center expansion end up hurting economic growth and smaller AI businesses, as well as companies such as Anthropic and OpenAI? If regulation is going to happen, it's going to take a while.

One of the reasons that serious AI regulation seems unlikely in the short term is that US President Donald Trump has loudly rejected calls for a slowing down of AI development, saying that "whoever wins AI, wins!" and calling AI safety fears a "hoax" concocted by the bosses at Anthropic, OpenAI, and SpaceX.

The President's point of view is that the guardrails that are already in place — essentially AI model oversight by the US government — are enough to keep AI from doing any harm. Trump will also be concerned about ceding ground to China, and the economic hit that may result in slower AI advances.

There's a huge amount of money invested in AI right now, from the companies themselves to the infrastructure that they rely on to run everything. If it takes longer to see returns on those investments — and OpenAI chief Sam Altman has already said his company's IPO is being pushed back over safety fears — then that could have significant economic impacts.

Assuming you're not using ChatGPT to develop bioweapons or hack into government mainframes, none of this noise around AI will affect the apps you use day-to-day at the moment. If you do try and do something malevolent with a chatbot, like writing computer viruses, you'll see there are many safety protocols in place to stop you.

What the AI companies are concerned about is keeping it that way. If they manage to coordinate some kind of slowing down and regulation, then you might see apps like Claude or Gemini updated less often. At the moment, new AI models are appearing on a regular basis, but that won't necessarily continue to be the case.

One possibility is that agentic AI, which is AI that's able to carry out tasks and program itself, becomes more tightly controlled even as other aspects of the technology improve.

What's certain is that this discussion is a long way from being over, and that those who know best say that we're at a "crunch time for humanity".

If you've been watching Reacher season 4 on Prime Video and think that it's been missing something that you can't put your finger on, never fear: highly-anticipated spinoff Neagley is finally here.

Maria Sten reprises her role as the titular character following her last on-screen appearance in Reacher season 3, and is now working as a private investigator in Chicago. As the official synopsis reveals, she quickly learns that a beloved friend from her past has been killed in a suspicious accident and becomes hell-bent on justice.

But, what does that actually mean? The creative team behind the spin-off are in a unique position: create an original story that can sit adjacent to an existing IP with a rich source material — that being, Lee Child's original Jack Reacher book series. Creative license is there to be taken and, in a way, it has been.

Having said that, creator Nicholas Wootton has confirmed that, across its eight episodes, all of which drop on Prime Video on September 16, we get to the bottom of Neagley's unsolved fear of being touched by, well, anyone. And, as Wootoon confirms, you've got your fellow fans to thank for getting an answer to this lingering conundrum.

"We're given the vibe of Reacher as a show, which is very specific, grounded, and real," Wootton exclusively told me ahead of release. "And, there's this one nugget of information that we get about Neagley, which is that she cannot touch people. We've no idea why and no clue about her backstory.

"To me, it was like 'This is our sandbox, and this is the one thing we can play with.' We built this season out from there. We wanted it to be a detective show like Reacher — there's going to be a mystery — but if it's successful in this first season, it's ultimately telling Neagley's story. The best first season of any show asks what we're learning about our main character, so if we can tell that through a detective story, we've got it."

As fans will find out over the course of the series, Neagley's family life and personal story is threaded through each episode. No spoilers here, but you're guaranteed to get answers as to why she has haphephobia... and they're pretty sobering.

The fact that we're finding any of this out to begin with, though, is all thanks to the fans. "She's such a popular character that is was clear that if a Reacher spinoff was being done, this was the direction that would be supported," Wootton explained. "To say that the fans have had a hand in it is absolutely true."

"I've really felt the love from the fans. Kudos to Amazon for plotting this all out... we shot Neagley before Reacher season 4 was filmed," lead star Sten added. "But, I love that they've decided to air them this way. We know and love Reacher, so it only made sense for Neagley to air right after. For the people who have missed Neagley, there's a lot more coming."

• Morefine MNAS-X1 stacks four hard drives directly above the processor to escape heat
• The MNAS-X1 packs an 800W power supply into a 15.6L chassis
• Four spring-latch drive bays let users swap storage without any tools

Morefine has released an AI workstation PC that stacks its storage drives on top of the chassis instead of beside the processor.

The MNAS-X1 arrives with a four-bay hard drive tray fixed directly above the motherboard rather than tucked beside it.

This unusual layout follows a fairly straightforward principle, since warm air naturally rises while cool air tends to settle lower, keeping the drives clear of the hottest components below them.

The chassis measures 218mm by 193mm by 372mm, giving a 15.6L footprint wrapped in a white aluminum alloy shell over a steel frame.

A glossy purple panel breaks up the front, and vents line the left, right, rear, and top surfaces for airflow.

Each of the four 3.5-inch bays has a spring latch release, letting drives slide out without tools once the side button is pressed.

Connectivity runs to one USB4 port, several USB-A ports, dual 10G RJ45 network jacks, an SD card slot, and a single HDMI 2.0 output for 4K at 60Hz.

Inside sits an AMD Ryzen 7 Pro 8845HS chip, an 8-core, 16-thread Zen4 chip with a 3.8GHz base clock and a 5.1GHz boost, paired with a Radeon 780M graphics core rated at 2700MHz.

An onboard XDNA NPU adds 16 TOPS on its own and 38 TOPS when the CPU, GPU, and NPU work together.

Two DDR5 SO-DIMM slots hold up to 96GB, three M.2 2280 slots take PCIe 4.0 x4 SSDs at 4TB each, and the four SATA bays add up to 30TB apiece for a combined ceiling near 132TB.

The three-slot PCIe 4.0 x16 opening accepts cards up to 330mm long and 66mm thick, backed by an 800W CRPS power supply built to the 2.0 specification with an 80 PLUS Platinum rating.

Fitted with 64GB of RAM, a 500GB M.2 SSD, and an RTX 5070 with 12GB of VRAM running in x8 mode, a review unit posted a Cinebench R23 score of 16271 pts multi-core and 1794 pts single-core, plus a PCMark 10 total of 9966.

Local language model testing through ONNX Runtime returned scores of 3511, 3714, 3163, and 3408 across the Phi 3.5, Mistral, Llama 3.1, and Llama 2 benchmarks, respectively.

That Cinebench multi-core figure lands in roughly the same range as an eight-core desktop chip like the Ryzen 7 5800X, while the PCMark total edges past what most mid-range gaming laptops score.

During a 21-minute full-load stress run, CPU power held at 54W with an average die temperature of 86℃, while the GPU drew 250W and averaged 80℃.

Morefine has not confirmed pricing for the MNAS-X1, and the unit is currently listed for inquiry only through its regional distributor.

Stacking the storage above rather than beside the board is a small design choice, but it turns a routine barebones NAS build into something closer to a hybrid server.

Whether that trade-off proves practical will likely come down to how much heat a fully loaded GPU pushes upward once the drive bay sits directly in its path.

Via HKEPC Hardware (originally in Chinese)

• Huawei's new Kirin 9050 Pro is being called its "most powerful" all-Chinese chip tech giant has ever made
• Huawei is claiming a 55% density gain by 'folding' its chip in half and stacking it instead of shrinking the actual die
• The chip still uses US hardware and foreign technology to manufacture, including ASML scanners and pre-2022 Applied Materials and Lam Research equipment

Huawei has introduced the Kirin 9050 Pro, a mobile processor that it says is fully free of American technology.

The chip powers the Mate XT 2, the third device in the company's tri-fold line, which starts at 19,999 yuan (~$2,980), and went on sale in China on September 12 2026.

The South China Morning Post noted Huawei is reducing reliance on advanced foreign semiconductor equipment, and this might be what the state-backed AI and chip giant wanted to highlight above all else, even as it refused to mention sanctions anywhere in its own press materials.

Richard Yu, chairman of Huawei's consumer business group, said on stage that the Kirin 9050 Pro is the company's "most powerful Kirin chip ever" and, based on the performance figures, touts a 42% overall improvement over the last-gen Mate XTs' Kirin 9020.

At the chip level, it claims 24% higher peak single-core and 52% better multi-core throughput against the Kirin 9030 Pro, plus a 142% jump in graphics rendering. The nine-core LinxiCore CPU peaks at 3.1GHz, the Maleoon GPU adds hardware ray tracing, and the Da Vinci NPU runs a 30-billion-parameter mixture-of-experts model on device.

The performance figures alone aren't the only interesting thing about Huawei's new silicon offering; the Kirin 9050 Pro is the first commercial silicon built on LogicFolding, a technique Huawei semiconductor chief He Tingbo presented at IEEE ISCAS in Shanghai in May and expanded on in a signed paper.

Rather than shrinking transistors, LogicFolding splits logic across two active silicon layers bonded face to face, with roughly 50 million vertical interconnects at a pitch of about 1.5 micrometers. Because signals travel shorter distances, Huawei says it could lower operating voltage while raising density.

However, all of these figures (including performance) come directly from Huawei, with no independent verification or peer-reviewed studies. More interestingly, while the Kirin is considerably less dependent on US or Western tech, it doesn't remove it from the equation entirely, and you still have to look closely at the bill of materials for receipts.

Huawei did not announce a process node for the Kirin 9050 Pro, and that silence has been deafening in more ways than one. SemiAnalysis tore down the predecessor in June 2026 and found SMIC's third-generation 7nm process, N+3, reaching about 113 million transistors per square millimeter with no extreme ultraviolet lithography anywhere in the flow.

SMIC's process, however, still leverages ASML immersion scanners. Bloomberg reported in 2024 that SMIC built the earlier Kirin 9000s using equipment acquired from Applied Materials and Lam Research before the October 2022 export rules closed that door, and the same is still in play here.

Despite this, Huawei has come a long way since its phones were declared America-free seven years ago. A UBS and Fomalhaut Techno Solutions analysis seen by the Wall Street Journal in late 2019 found the Mate 30 Pro contained no American parts, and Huawei cybersecurity officer John Suffolk told the paper that "all of our 5G is now America-free." The Kirin 990 inside that phone, however, was fabricated by TSMC on an EUV node.

Huawei's approach, backed by the Chinese state, to essentially rid itself of US influence in the fabrication end of the spectrum remains more of an ongoing struggle than one which it has already achieved, even as it continues to push ahead in certain disciplines and erode a performance gap that industry players earlier thought would take years longer than it is now projected to.

• Watch Run Hide Fight: Infidels on Daily Wire
• Traveling? Watch streaming services from anywhere with our no.1 VPN (SAVE 75%)

Jonathan Majors (Creed III) is back — and armed for battle in Run Hide Fight: Infidels, a provocative new action thriller about a battle-hardened Delta Force veteran thrown into a deadly campus siege.

Billed as a 'standalone follow-up' to 2020's Run Hide Fight, which starred Isabel May as teenager Zoe Hull, who fights back when gunmen attack her high school, Infidels aims to reboot the franchise.

The original, which premiered at the Venice Film Festival, received a sharply divided reception: it currently has 30% from critics but 93% from audiences on Rotten Tomatoes (1,000+ audience ratings).

Expect to see plenty of social media posts about this provocative movie. If you want to judge for yourself, here's how to watch Run Hide Fight: Infidels online from anywhere.

• TV tips email: Never be stuck for anything to watch again

Run Hide Fight: Infidels premieres exclusively on DailyWire Plus on Wednesday, September 16.

The streaming service costs $15 per month (cancel anytime). Or choose an annual plan for 40% off.

Traveling outside the States? You may need to use a VPN to unblock DailyWire Plus when abroad. It's also a great way to stream content privately.

If you're out of the country but keen to stream Run Hide Fight: Infidels (2026) you might find local restrictions in place.

Luckily, there’s an easy solution. Downloading a VPN will allow visit your usual websites and streaming services no matter where you are. It's a simple bit of software that changes your IP address, meaning that you can access on-demand content or live TV as if you were back at home.

Get 75% off Nord VPN with this deal to stream safely and privately...

NordVPN – get the world's best VPN
We regularly review all the biggest and best VPN providers and NordVPN is our #1 choice. It unblocked every streaming service in testing and it's very straightforward to use. Speed, security and 24/7 support available if you need – it's got it all.

The best value plan is the two-year deal which sets the price at $3.99 per month, and includes an extra 3 months absolutely FREE. There's also an all-important 30-day no-quibble refund if you decide it's not for you.

- So, try NordVPN 100% risk-free for 30 daysVIEW DEAL ON

Yes, Run Hide Fight: Infidels is also streaming on Daily Wire in Canada, the UK and Australia.

If you are worried online privacy or security, we highly recommend that you invest in a VPN. Our favorite, and the best value for money, is NordVPN (Save 75% and try for 30 days).

• Jonathan Majors — Clayton “Clay” Saunders / Captain Saunders
• JC Kilcoyne
• Stephen Hailo — Ibrahim Asad Melod
• Tyler Aser — Lizzy Dupree
• James Bingham — P.J.
• David Lucas
• Charlotte Delaney Riggs — Farrah Abel
• Jason Gray-Stanford
• Myles Clohessy
• Chase Ramsey — Hermann
• Sharon Johal
• Paulina Alvarez
• Erin Davie
• Siaka Massaquoi
• Travis Mills

VPN services are evaluated and tested by us in view of legal recreational use. For example:a) Access to services from other countries, (subject to the terms and conditions of that service).b) Safeguarding your online security and making your online privacy more robust when abroad.Future plc does not support nor condone the illegal or malicious use of VPN services. We do not endorse nor approve of consuming pirated content that is paid-for.

• Nvidia CEO Jensen Huang and Anthropic CEO Dario Amodei discuss AI at Dreamforce 2026
• Amodei says the pace of progress had surprised Anthropic
• Huang calls for greater safety, but hails the advantages AI can bring

Some of the world's leading AI minds have shared their opinions on the future of the technology, revealing some intriguinly diverging viewpoints.

Nvidia CEO Jensen Huang and Anthropic CEO Dario Amodei were both guest appearances during the opening keynote of Dreamforce 2026, where they were both quizzed by Salesforce CEO Marc Benioff on their opinions and views on the future of the technology.

The pair had some interestingly differing viewpoints, showing the trouble facing not just internal and external stakeholders, but also lawmakers and regulators.

First up was Amodei, whose Claude platform has become increasingly partnered with Salesforce in recent months, culminating in the recent launch of Claudeforce.

Asked by Benioff about the state of the AI market right now, Amodei likened it to the motor industry, where safety records can be conflated, but companies can learn from each other - and he also seems keen to open things up internationally to set standards.

“I think that’s the way to lead the industry forward, to set an example, to say that everyone can always be better,” he pointed out.

Benioff also asked Amodei about the biggest surprises has had while developing his technology over the last decade.

The Anthropic head replied the "pace of the progress" - not just in the development of AI technology, but the economic impact of it too - noting his company didn't appreciate how quickly certain companies would grow and become central to everyday life.

Finally, Benioff asked what actions Amodei would like the public to take when using AI.

Amodei replied that even if development of AI technology was frozen right now, we'd only be using maybe 5-10% of the possible value we could get.

Making use of all the information with platforms such as Claude could open up new areas of business - "there's so much diffusion left," Amodei says.

Huang was much more positive in his appearance, leading Benioff in a walkabout around the keynote theatre as the pair discussed AI.

"We're going through a new industrial revolution," Huang pointed out,

However he did urge caution in some areas, particularly around safety.

“Safety is paramount in a lot of ways. It’s job one,” Huang said. “However, safety is an engineering problem.”

Speed and safety are not mutually exclusive, Huang noted, pointing out that companies should pace themselves until they are sure they are releasing something that the market will appreciate.

“It’s a false choice,” Huang told Benioff. “You could definitely have both at the same time.”

When it comes to model protections, Huang said that companies should, “run as fast as you can...but if you feel at any given point in time the company’s out of control or the product’s not going to be safe, take a pause and make sure you get it right.”

Wrapping up, Benioff asked Huang about his vision for the next few years - and what he wants to see from Salesforce.

Huang says he wants to "Agentforce every company", outlining an aim of "really deploying AI".

"You don't want to be left behind, this is too important of a technology revolution," Huang noted, "if it doesn't work for you straight away, don't give up on it."

"The most important thing - don't get left behind...the sky's the limit for us."