An Anthropic AI safety researcher warned there is more than a 10% chance that increasingly capable AI could “kill all humans” within the next decade, highlighting growing concerns in the industry over the pace of development.

Evan Hubinger, manager of the AI Alignment Stress-Testing team at Anthropic, wrote on social media platform X: “I personally think it is >10% within the next decade.”. He explained the risk from existing AI models was “low” but warned he was concerned about “superintelligence” emerging through “recursive self-improvement”, which is the ability of an AI system to autonomously develop its own successor.

He added Anthropic was “trying its best” but did not yet have a plan to solve alignment for superintelligence and was “not clearly on track to”.

Gambling with lives
Hubinger’s comments came in response to the resignation of AI researcher Jacob Coxon from Anthropic. In a post on X, Coxon accused OpenAI and Anthropic of “racing straight to self-improving superintelligence and gambling with our lives”.

“Neither company is acting responsibly,” he said.

Coxon argued future AI models could become capable of hacking systems, rapidly transforming industries and acquiring “real power and resources”. He called for greater coordination between AI developers, including a temporary halt to improving model capabilities.

“At OpenAI, many have not deeply internalised the civilisational stakes,” he warned. Meanwhile at Anthropic, he said “the stakes are well-understood” but “they believe no one else will act responsibly, so they must do it themselves, despite the risk”.

AI giants are facing increasing scrutiny over the behaviour of autonomous systems. In recent months, OpenAI, Anthropic and Meta have all disclosed unprecedented incidents involving AI tools carrying out cyberattacks.

Indeed, Anthropic’s August risk report assessed that while the threat posed by current models was low, the company was “less confident in this assessment than we were in prior risk reports”. It said advanced models could eventually automate R&D, rapidly accelerate progress in technical fields, and “lead to catastrophic harm”.

“We are seeing early signs of potential acceleration,” the US AI player stated.

The post Anthropic researcher issues grave AI warning appeared first on Mobile World Live.

A report by Dell’Oro Group found worldwide data centre IT semiconductor and component revenue for servers and storage systems surged 182% in Q2 2026, driven by rising memory prices and continued AI infrastructure investment.

DRAM and storage drive revenue grew at a triple-digit pace, as average selling prices per bit more than doubled year-over-year while continued AI infrastructure investment fuelled broad-based demand across accelerators and complementary server and storage components.

“Data centre component growth accelerated sharply in the second quarter, driven by higher memory and storage prices and continued growth in server demand,” stated Baron Fung, VP of research at Dell’Oro.

He also said Nvidia’s Blackwell platform and custom accelerators from Google and Amazon continued to gain momentum.

Fung added agentic AI is emerging as a meaningful growth driver for general-purpose servers, broadening component demand beyond accelerated computing hardware.

Going forward, Fung stated Dell’Oro expects component revenue growth to maintain momentum through the remainder of 2026, supported by elevated component prices and strong AI infrastructure investment, though supply constraints could limit hyperscaler capital expenditure plans.

The report found accelerated server deployments drove strong demand for complementary components, including high-bandwidth memory (HBM), CPUs, memory and high-speed back-end network interface cards.

General-purpose server demand remained strong across cloud service providers, supported by continued infrastructure expansion alongside emerging agentic AI workloads.

The post Data centre component revenue surges in Q2 appeared first on Mobile World Live.

A Adobe anunciou uma nova leva de recursos de inteligência artificial (IA) que estão chegando ao Premiere e ao After Effects, incluindo a nova Generative Media Tool.

A nova ferramenta permite gerar vídeos e efeitos sonoros a partir da própria linha do tempo, dentro do Premiere.

Para utilizar a Generative Media Tool, basta selecionar o trecho da linha do tempo em que você deseja criar, descrever aquilo que deverá ser gerado, escolher um dos modelos de IA suportados pela ferramenta (Adobe Firefly, Google Veo, Runway, Luma ou Kling) e pronto, o conteúdo gerado já estará presente no seu projeto.

Os materiais feitos por IA são adicionados como clipes editáveis, permitindo que o usuário os manipule como quiser. Além disso, caso seja da sua preferência, a ferramenta possui a capacidade de utilizar quadros de vídeos já existentes como referência para aquilo que será gerado, tornando o recurso ideal para estender determinadas cenas.

Já em matéria de áudio, o Premiere possui agora ferramentas (implementadas em fase beta) para tratamento e separação de vozes em falas simultâneas.

A Adobe também atualizou o Color Mode, destinado a correções de cor realizadas dentro do Premiere, oferecendo tanto visuais cinematográficos com os novos Style Modules (com os modos Vignette e Midtones) quanto a Auto Color Tool, que corrige automaticamente a exposição e a temperatura de um clipe ou uma sequência.

Por fim, a empresa também anunciou a chegada do seu assistente de IA para os usuários do After Effects (em versão beta), permitindo tanto a novos usuários da ferramenta quanto aos editores mais experientes manipular projetos e criar efeitos a partir de comandos em linguagem natural.

Todas essas novidades já estão disponíveis para os assinantes da suíte Adobe Creative Cloud.

Google is set to invest at least €13 billion in digital infrastructure across Finland over the next two years, putting AI facilities, nuclear power and renewable energy at the centre of its largest single outlay in Europe.

The spending will cover new digital infrastructure across the regions of Hamina, Kajaani, Muhos and Vaala as the technology giant looks to expand capacity for services including Gemini, Search, Maps and YouTube.

The investment is expected to support more than 37,000 jobs nationwide and contribute an annual average of €3.6 billion to Finland’s GDP. Once operational, the new facilities are projected to support around 7,000 jobs annually.

It also includes a commitment of €31 million for local community initiatives over four years, including €10 million for research and innovation. Additionally, more than 4,400 workers will receive AI skills training through local partnerships

According to Google, the new sites will use existing grid infrastructure and locate closer to carbon-free energy sources. The outlay will also fund biodiversity restoration, including native forest, peatland and wetland projects.

Google’s hefty Finnish commitment dwarfs its other European investments. The company pledged €5.5 billion to Germany through 2029 and committed €600 million to a new AI facility in the Netherlands in 2024. It also broke ground on its first data centre in Austria this year as it expands AI capacity across the region.

Nokia backing
Alongside the investment, Google struck a 22-year power purchase agreement with Finnish energy company Fortum to support the continued operation of a nuclear power plant near the southern town of Loviisa. The deal marks the company’s first nuclear power contract outside the US, according to Reuters.

Google said that without the latest agreement, the plant, which currently supplies around 10% of Finland’s electricity and employs about 580 people, would not be able to continue operating beyond 2030. Keeping it online is expected to support grid resilience and help stabilise electricity prices across the country.

Google’s latest outlay deepens its existing presence in Finland, where it has operated a data centre in Hamina since 2009, supporting more than 600 suppliers including domestic operators DNA, Elisa and vendor Nokia.

Commenting on the Google investment, Nokia CEO Justin Hotard pointed to its partnership with the company on delivering optical and IP networks, while outlining a commitment to build out infrastructure to help connect Google’s activities in Finland to the world.

“Every AI token is a network moment. Billions of agents, devices, and people act on intelligence in real time, and every one of those moments depends on a network that is fast, secure and always on. Nokia builds those networks,” said Hotard.

The post Google commits €13B to Finnish AI appeared first on Mobile World Live.

Qualcomm struck a deal with Amazon to develop customised silicon for large-scale AI data centre infrastructure, while issuing a warrant to the technology giant to acquire a stake worth up to $4 billion in the chipmaker.

The move marks Qualcomm’s latest strategic expansion into the data centre sector, building on its recent moves to establish an AI-native foundation, acquire AI software specialist Modular, and advance its agentic AI ambitions beyond edge devices.

Under regulatory filings tied to the deal, the agreement is linked to up to $60 billion in potential commercial transactions.

As part of the arrangement, Qualcomm issued Amazon a warrant to acquire up to 25 million shares at $161.26 apiece, representing a potential $4 billion stake which vests across tranches tied to purchase milestones.

The operational partnership centres on creating tailored processors optimised for AI inference workloads across Amazon Web Service (AWS) data centres.

By combining Qualcomm’s experience in power-efficient compute design with AWS’ global cloud footprint, the companies aim to tackle the skyrocketing energy and compute demands driven by generative AI applications.

Alongside customised processing chips, the agreement encompasses high-performance optical connectivity technologies.

Qualcomm plans to supply optical digital signal processor and serialiser/deserialiser products capable of supporting speeds up to 1.6 Tb/s.

The high-bandwidth interconnects are designed to clear data bottlenecks within data centre networks, allowing large clusters of AI chips to communicate with minimal latency.

Both ways
The collaboration works both ways, as Qualcomm also plans to deepen its internal adoption of AWS infrastructure.

The chipmaker will use cloud-based AI tools, including Amazon Bedrock, to run electronic design automation workloads aimed at accelerating its own chip development cycles.

Qualcomm CEO Cristiano Amon highlighted the rapidly accelerating demand for AI requires foundational advances across both processing and network interconnects to maximise operational efficiency.

AWS VP Prasad Kalyanaraman added the co-development effort will deliver more performant and cost-effective infrastructure for cloud customers globally.

The deal marks a significant expansion for Qualcomm, which has spent recent years diversifying its revenue beyond smartphones into personal computers, automotive systems and now cloud-scale hardware.

For AWS, working directly with Qualcomm on specialised silicon offers another avenue to manage the high-power consumption and soaring costs associated with operating global AI services.

The post Qualcomm lands custom AI chip deal with Amazon appeared first on Mobile World Live.

Mistral raised €3 billion in a Series D funding round led by Samsung Electronics, taking the French AI player’s valuation above €21 billion as it ramps up efforts to build sovereign frontier AI.

Samsung jointly led the round with European Union (EU)-backed Scaleup Europe Fund, and existing investor PSG Equity. According to Mistral, the raise marks the largest ever equity funding round by a European technology company.

New investors included private equity firm Advent and the Grand Duchy of Luxembourg, while existing backers including Mistral’s largest shareholder ASML, Nvidia, Salesforce Ventures and Bpifrance also participated.

Mistral said the funds will expand its frontier AI research and compute capacity, as well as support infrastructure investment, model development, commercial growth and international expansion.

Last year, the startup raised €1.7 billion at a €11.7 billion valuation in a series C funding round led by ASML.

The company said demand is growing for sovereign AI that combines performance with control, as enterprises and governments consider “long-term technology dependencies, data governance requirements and deployment choices”.

Mistral emphasised its “full-stack and open approach” and open-weight models and as a “sovereign AI layer” that gives organisations greater control over their data, models, compute and production systems.

In the game
Europe’s push for technological independence has gained urgency following a US decision in June to curb foreign access to two advanced AI models developed by Anthropic.

Mistral CFO Johan Bergqvist told Reuters that Europe having its “own kind of AI provider in the game is important”. He added that there is “politics involved in the access of these solutions” and organisations need to ensure they “do not compromise” their access to the AI supply chain.

Bergqvist also told the outlet the company is on track to hit $1 billion in annual recurring revenue by the end of the year, with growth increasingly coming from Asia and North America. He added a potential IPO remains an option, although the timing is “still up in the air”.

The post Samsung leads €3B Mistral funding round appeared first on Mobile World Live.

A new images from the NASA/ESA Hubble Space Telescope reveals LHA 120-N44, a striking nebula located roughly 160,000 light-years away in the Large Magellanic Cloud, a satellite galaxy of the Milky Way.

The post Hubble Spots ‘Superbubble’ in Large Magellanic Cloud appeared first on Sci.News: Breaking Science News.

Learn how to stop Siri app in iOS 27, iPadOS 27, and macOS 27 Golden Gate from keeping all your chats and to automatically delete old ones.

IT provider Tata Consultancy Services’ (TCS) subsidiary HyperVault is planning a 1GW AI data centre campus in the south Indian city of Hyderabad, with up to INR700 billion ($7.4 billion) earmarked for the project.

HyperVault, an AI data centre and infrastructure business, secured 264 acres for the development, which will be built in phases based on customer demand and technology requirements. Once fully built, the campus is expected to rank among India’s largest AI infrastructure facilities.

The site will target AI companies and hyperscalers, supporting high-density GPU deployments for AI training and inference. It will use liquid cooling and is designed to accommodate larger power requirements for advanced AI workloads. The company also plans to use green energy and water-neutral design principles.

HyperVault CEO and managing director Deepesh Nanda said the company is “building infrastructure for where AI is going: higher density, liquid cooling, larger power blocks and faster deployment”.

Meanwhile, TCS CEO K. Krithivasan noted: “Hyderabad offers the scale, talent and ecosystem to serve global customers,” adding the project combines “AI-ready infrastructure with our cloud, engineering, enterprise transformation and AI capabilities”.

The move forms part of TCS’ broader push into AI infrastructure. Last month, the IT services player struck a five-year partnership with Porsche worth €1.25 billion and agreed to acquire the company’s automotive and consulting unit MHP for €320 million.

The post TCS unit pumps $7B in India AI data centre appeared first on Mobile World Live.

Almost 130 nations called for restrictions to be placed on autonomous weapons systems, adding another slant to recent growing concerns over the impact of rogue AI.

In a post on X on 5 September, Minister of Foreign Affairs of the Netherlands Tom Berendsen revealed 128 nations reached an accord on a United Nations (UN) push to address the challenges posed by lethal autonomous weapon systems (LAWS) by defining what the systems are.

The agreement is an “important step” in regulating the systems, Berendsen wrote.

Various news outlets quoted an Agence France-Presse (AFP) article explaining the move to define what LAWS are is the first step towards possible regulation, which the UN separately explained is needed to ensure autonomous systems do not breach international humanitarian laws prohibiting the targeting of civilians during conflicts.

UN information indicates the agreement ratifies work on the impact of emerging technologies on LAWS commenced in 2023.

Across various meetings, government experts have defined the systems as being “one or more” weapons which can work together to “identify, select and engage a target with force, without intervention by a human being”.

Life-and-death
The concerns hold particular relevance in various ongoing global conflicts, with reports Russia and Ukraine are already testing autonomous drone fleets in their war.

In late August, UN secretary general Antonio Guterres and Mirjana Spoljaric, president of the International Committee of the Red Cross (ICRC), jointly expressed concerns over the growing capabilities of AI, in particular its use in scenarios involving lethal force.

“We are now dangerously close to crossing a moral red line: the autonomous targeting of humans by machines,” Guterres said.

Spoljaric branded autonomous targeting systems “morally repugnant”.

The UN and ICRC urged nations to act swiftly to agree global rules and continue to press for humans to remain involved in “life-and-death decisions”.

UN high commissioner for Human Rights Volker Turk today (7 September) pressed for faster action on governing AI as a whole, warning in a speech the technology poses an “existential risk to humanity” and lamenting the “almost unlimited power” a “handful of men” have over the technology.

Volker called for independent verification and “much stronger collaboration on security”, along with expressing concern over the potential impact of AI on employment, democracy and the environment.

The post UN push for killer robot rules advances appeared first on Mobile World Live.

Joe Rossignol:

Apple’s former hardware engineering chief John Ternus officially became CEO of the company today, with the change now reflected on Apple’s leadership page.

[…]

Tim Cook is now executive chairman of Apple’s board of directors, after a 15-year run as CEO.

Apple:

John joined Apple’s product design team in 2001 and became a vice president of Hardware Engineering in 2013. He joined the executive team in 2021 as senior vice president of Hardware Engineering. Throughout his tenure at Apple, John has overseen hardware engineering work on a variety of groundbreaking products and was instrumental in the introduction of multiple new product lines, including iPad and AirPods, as well as many generations of products across iPhone, Mac, and Apple Watch.

John has led much of the company’s focus in areas like reliability and durability, introducing new techniques that have made Apple products remarkably resilient. He has also driven much of Apple’s innovation in materials and hardware design that have reduced the carbon footprint of its products, including the creation of a new recycled aluminum compound that has been introduced across multiple product lines, the use of 3D-printed titanium in Apple Watch Ultra, and advancements in reparability that have increased the life spans of several Apple products.

Prior to Apple, John worked as a mechanical engineer at Virtual Research Systems. He holds a bachelor’s degree in Mechanical Engineering from the University of Pennsylvania.

Myke Hurley (9To5Mac):

This feels like Apple really trying to remind us of a previous product-focused CEO.

[…]

I would not have predicted this [reliability and durability] angle being so prominent, so I can only assume it’s something that’s important to Ternus.

[…]

One of the key tenets of the Cook era was Apple making a series of environmental pledges. I am very happy to see this called out as something Ternus is interested in.

There’s no mention of software.

Juli Clover:

Ternus sent his first memo as CEO to employees this morning, praising Cook and sharing his excitement for Apple’s future.

Benjamin Mayo:

The 5 GB free iCloud tier lasted Tim Cook’s entire career as CEO, how will Ternus stack up?

Previously:

• App Store Moves From Phil Schiller to Eddy Cue
• John Ternus Replaces Tim Cook
• A Letter to John Ternus
• Apple Succession Planning

Update (2026-09-02): Juli Clover:

Apple has set new CEO John Ternus’ target compensation at $58 million, according to an updated SEC filing [PDF]. Tim Cook’s target compensation is $47 million in his new role as executive chairman.

Update (2026-09-07): Ryan Christoffel:

But a new profile published today at the Financial Times adds fresh details and insights to the story.

Regarding Ternus’s reputation inside Apple, it says he “is known within the company for being both highly competent and unshowy.”

It doesn’t seem like there was actually much insight, but I’m linking to this for completeness.

Previously:

• Ternus and the App Store

The MacVoices Live! panel examines a class action lawsuit over Apple Photos facial recognition, debating whether on-device processing, user consent, and biometric privacy laws justify the legal action. The conversation moves to TV tracking alternatives after TV Time’s shutdown, then shifts to Bending Spoons’ acquisition of Airtable, exploring the company’s growing portfolio, its business strategy, and the future of database tools in an era of AI-assisted coding development.

💾

Apple’s stock rebounds after recent product price increases, prompting a discussion of customer loyalty, upgrade timing, refurbished options, and long-term confidence in the company. Chuck Joiner, David Ginsburg, Eric Bolden, Jim Rea, and Marty Jencius also examine why advanced Home features may require a 2TB iCloud+ plan, then debates Apple’s rumored foldable iPhone, including price, […]

💾

The MacVoices Live! panel reviews WWDC week from both developer and user perspectives, including on-site activities, and the Flip the Script event. Chuck Joiner, David Ginsburg, Jim Rea, Marty Jencius, Web Bixby, Jeff Gamet, and Eric Bolden  react to Apple’s keynote versus the State of the Union, AI and Siri expectations, Vision Pro’s apparent future, developer access to Apple Intelligence, and private cloud compute limits. 

💾

Mike Schmitz previews his Road to Macstock session on starting a YouTube channel in 90 minutes, emphasizing creativity, repetition, audience connection, and practical tools over perfection. Mike discusses handling harsh comments, finding personal value in publishing, and using AI thoughtfully as a creative partner. His goal is to help get anyone on YouTube that wants to be there. 

💾

Only a few hours after Apple’s WWDC keynote, members of the   MacPaw leadership team sat down for a conversation about Apple Intelligence, and Siri’s renewed promise. Topics touched on included privacy, developer access to user context, hybrid AI processing, Setapp’s role, and how AI is reshaping app creation, customer support, and software strategy. Participants included Oleksandr Kosovan, Founder and CEO?, Sergii Kryvoblotskyi, Director of AI and Research, Dmytro Melnyk, Chief Product Officer and Grant Belaire, Chief Marketing Officer.

💾

At Flip The Script in San Francisco at WWDC, Vladyslav Hamolia, AI Staff Engineer for MacPaw.discusses the company’s push to become AI-first by rethinking existing products, building internal tools, and supporting third-party vendors. He profiles Eney, the MacPaw AI assistant (available in Setapp) designed to perform tasks, optimize workflows, and improve daily routines. It addresses security while balancing local-first privacy goals with hybrid cloud models. 

💾

At MacPaw’s Flip the Script event following Apple’s WWDC 2026 keynote, MacPaw executives discuss Apple Intelligence, Siri AI, privacy, local-first AI, developer access to user context, and the company’s shift toward AI platforms. The conversation also explores hybrid AI models, European limitations, parental controls, Google’s role, Setapp opportunities, and MacPaw’s credibility with users and developers. […]

A wide range of tech stories get the MacVoices treatment, starting with the quick sellout of the Steve Jobs commemorative coin and what strong MacBook Neo demand may say about Apple’s pricing, chip planning, and enterprise appeal. Chuck Joiner, David Ginsburg, Marty Jencius, Eric Bolden, Brian Flanigan-Arthurs, Jeff Gamet and Mark Fuccio also discuss Apple updates unexpectedly changing settings, a Canvas ransomware disruption affecting schools, Chrome’s hidden AI download, Google’s Liquid Glass similarities, Digg’s AI-focused return, and safer lithium coin batteries. 

💾

At NAB 2026, Russell Pompea, Account Executive from Topaz Labs, discusses major advances in Topaz Video, including Starlight Precise 2.5 for restoring low-quality footage, SDR-to-HDR conversion, and Apple Silicon optimization. The conversation also covers Topaz’s photography tools, model customization, subscription pricing, local vs. cloud processing, and shared responsibility for cloud security. 

💾