Anthropic expanded the reach of its Mythos AI model to an additional 150 companies across 15 countries but stated each will need to meet its security requirements before they gain access.

Anthropic introduced its Claude Mythos model on 7 April, under the auspices of its Project Glasswing to a limited number of technology companies including Amazon Web Services, Apple, T-Mobile US, AT&T, Nvidia and Google, instead of making it publicly available.

The company stated the new cohort features industries which were underrepresented in the first batch. It now includes power grids, water systems, healthcare networks, communications providers, and hardware manufacturers.

Anthropic stated for most of the Project Glasswing partners, a successful cyberattack on their codebases could affect more than 100 million people.

It also noted many of the new partners are vendors, companies or nonprofits that maintain codebases which are relied upon by numerous organisations around the world, including governments.

The company expects within six-to-12 months, many other AI developers will have models comparable to Mythos Preview and stated, “they could release them without safeguards that prevent misuse”.

Results from the first cohort are already in. Project Glasswing partners have collectively surfaced more than 10,000 high-or critical-severity security vulnerabilities in the first few weeks.

The AI player stated the bottleneck in cybersecurity is now verifying, disclosing, and patching the large numbers of vulnerabilities which Mythos-class models can surface.

It noted many of Project Glasswing’s partners now use the model to write patches, as well as for pre-release checks which prevent vulnerabilities from appearing in the first place.

The expansion came a day after the Anthropic stated it will start offering Mythos access to the European Union’s cybersecurity division.

It also confidentially filed its initial public offering prospectus with the US Securities and Exchange Commission ahead of rival OpenAI.

The post Anthropic expands Mythos access to 150 new companies appeared first on Mobile World Live.

AI player Anthropic confidentially submitted paperwork for its proposed initial public listing ahead of rival OpenAI, while also giving the European Union’s cybersecurity body preliminary access to its Mythos AI tool.

The draft registration statement submitted to the US Securities and Exchange Commission gives the company the option to go public after the agency completes its review.

Anthropic stated the number of shares to be offered and the price have not yet been set.

News of the IPO move came the same day (1 June) Bloomberg reported Anthropic will give ENISA, the European Union’s cybersecurity agency, access to Mythos through Project Glasswing, an initiative which allows organisations to test Mythos’ capabilities before a wider release.

There are growing concerns among governments over the security implications of Mythos, which Anthropic released to some private companies in April.

Anthropic communicated the decision to the European Commission over the weekend.

EC spokesperson Thomas Regnier confirmed the development to Mobile World Live (MWL) followed several weeks of productive discussions.

 “We welcome the latest developments on potential future access,” he said. “This is the result of the Commission’s strong bilateral cooperation and engagement with Anthropic, a leading frontier AI company.”

The EC was careful to frame the moment not as a resolution but as a starting point to work with the US administration, Anthropic and additional AI companies such as OpenAI.

“This is a shared challenge, and we are intensifying our discussions with like-minded partners, including the United States,” Regnier said.

The plan is for ENISA to join Project Glasswing, the coalition Anthropic announced in April which includes Amazon, Apple, AT&T, T-Mobile US, Microsoft, Google, CrowdStrike, Nvidia and Palo Alto Networks, among others.

The post Anthropic confidentially files for IPO appeared first on Mobile World Live.

SoftBank Group committed to an investment of up to €75 billion to bolster AI data centre infrastructure in France, with the first phase of the project set to deliver 3.1GW of capacity.

SoftBank announced the investment at the 2026 Choose France summit, hosted by President Emmanuel Macron, marking the Japanese company’s largest AI infrastructure investments in Europe.

It has committed an initial €45 billion investment in the Hauts-de-France region, providing 3.1GW of capacity to data centres in Dunkirk, Bosquel and Bouchain. SoftBank will also develop additional sites, “reinforcing the country’s role as a leading European hub for next-generation digital infrastructure”.

For the Dunkirk deployment, SoftBank partnered with Schneider Electric to accelerate its buildout, while developing a large-scale industrial production cluster.

The cluster at the Port of Dunkirk will be a “key industrial pillar” for the company’s AI infrastructure programme in France, including the build out of two facilities. One will be operated by SoftBank to manufacture enclosures, while the other will be operated by Schneider Electric to integrate data centre power modules.

The duo explained the partnership will combine SoftBank’s robotics and automation capabilities with Schneider’s industrial expertise and local supply chain network to support the deployment of next-generation AI data centres at scale.

The industrial cluster is also designed to support Dunkirk’s ambition to become a leading hub for robotics, advanced manufacturing and industrial innovation.

Masayoshi Son, chairman and CEO of SoftBank, said AI is entering a new era and countries that build infrastructure for this transformation “will shape the future of technology, industry and society”.

“SoftBank is proud to make this major commitment to France. With its industrial capabilities, talent base and national ambition, France is uniquely positioned to become a leading AI infrastructure hub in Europe.”

The company said it will also work with SB Energy and other strategic partners to deliver the projects.

The post SoftBank to splash up to €75B on France AI capacity appeared first on Mobile World Live.

Qualcomm CEO Cristiano Amon used his keynote at the annual Computex event in Taiwan to stake the company’s claim in the next phase of AI, arguing the technology will reshape demand for compute across devices, networks and data centres.

Amon described 2026 as the “year of the agent”, stating AI is moving from prompt-based interactions to autonomous systems capable of planning, reasoning and acting across smartphones, PCs, cars, robots and industrial equipment.

“Agents are not coming in the future. They’re already here,” he said, adding the shift is “changing a lot of the compute” and could generate “a lot of demand for new classes of devices and computing”, creating “one of the largest” upgrade cycles the industry has seen.

Amon said the smartphone will no longer sit alone at the nexus of the digital ecosystem. “Agents become the centre of your digital experience,” he stated, adding devices will increasingly become “endpoints for agents”.

Compute continuum
To this end, the executive laid out Qualcomm’s ambition to support the AI infrastructure transition. Amon pointed to the need for CPUs, GPUs, NPUs and connectivity designed to support AI workloads both on devices and in the cloud, stating the company can help scale AI compute from “sub-2 milliwatts” in devices such as earbuds to kilowatt-level systems in data centres.

He also stressed the engineering challenge around battery life and latency, noting devices must be able to support complex planning, reasoning and coordination. “I cannot emphasise enough the importance of power,” he said.

In addition, Amon framed 6G as a key part of the future AI architecture, noting it is the first wireless generation designed as an AI-native network connecting distributed, hybrid intelligence across devices and data centres.

During the event, the chief also unveiled Dragonfly, Qualcomm’s new data centre brand aimed at inference workloads. He said Qualcomm is already working with hyperscalers and global partners on deployments, adding the fresh brand will allow its portfolio to span “every single tier of the compute continuum”.

The post Qualcomm boss sets out agentic AI ambitions appeared first on Mobile World Live.

The European Union (EU) is pressing for deeper talks with the US administration over advanced AI models, and at the heart of the conversation is Anthropic’s Mythos.

There are growing concerns among governments over the security implications of Mythos, which Anthropic released to private companies in April.

Its release triggered an immediate wave of concern when it surfaced the model could identify tens of thousands of software vulnerabilities at a scale no previous system had demonstrated.

The AI player introduced its Mythos model on 7 April, under the auspices of Project Glasswing, to a limited number of technology companies including Amazon Web Services, Apple, Nvidia and Google.

Anthropic expects to bring Mythos-class models to all customers in the coming weeks.

Bloomberg previously reported the EU made limited progress in securing access to details of vulnerabilities Anthropic’s Mythos AI model could reveal.

European Commission spokesperson Thomas Regnier told Mobile World Live (MWL) the agency has had several meetings with Anthropic to understand the capability of the model, its implications for the cybersecurity of the EU and Anthropic’s plan around Project Glasswing.

“We will keep discussing with the company the cyber capabilities and risks of its latest model,” he stated.

CNBC reported Anthropic has yet to grant the EU, its AI office or any government organisations outside of the US, aside from the UK’s AI Security Institute, preview access to Mythos.

Since August 2025, the European Commission’s AI Office has held regular technical meetings with Anthropic tied to the General-Purpose AI Code of Practice, to which the company is a signatory.

A spokesperson for the EC noted Mythos is not a one-off as a “new wave of powerful models are coming to the market”.

The EC stated parallel progress is being made towards releasing OpenAI’s GPT-5.5-Cyber to trusted EU entities.

The EC spokesperson told MWL it is intensifying discussions with the US, “particularly on the most advanced AI models, including those with cyber capabilities”.

“Cybersecurity is a shared priority and we have agreed to mutually recognise our respective standards in this area,” the spokesperson stated.  “On EU side, we are also stepping up our cyber defences through targeted investments in AI and supercomputing.”

The post EU pushes for access to Anthropic model as fears grow appeared first on Mobile World Live.

NTT Docomo Global expanded its work with Accenture and AWS to build infrastructure for enterprise agentic AI focused on governance and ensuring trust in systems.

The collaboration is set to centre on further developing the NTT unit’s Universal Wallet Infrastructure (UWI), a platform developed with Accenture to manage digital identity, credentials, money and documents across different apps, wallets and services.

Under the latest pact, NTT will provide the UWI trust infrastructure layer, while Accenture will bring technology strategy, digital assets and product engineering. AWS will contribute cloud and AI services.

NTT stated the expanded work targets a growing governance gap as AI agents increasingly write and modify code across development environments. It argued traditional security and software supply chain approaches were not built to monitor autonomous systems operating continuously at scale.

The partners plan to embed identity, credential and policy controls into workflows, allowing AI actions to be verified, governed and audited. The focus is initially on software development, though the companies are eyeing broader enterprise applications.

The trio will also carry out joint go-to-market activities including customer workshops, product showcases and educational sessions.

NTT Docomo Global CEO Hiroki Kuriyama said “the next chapter of AI will depend on whether people, enterprises, and society can trust how intelligent systems behave and interact”.

AWS MD Asia Pacific, Japan and China Jaime Valles added customers want to move quickly with agentic AI, but need “trust and governance built in from day one”.

The post NTT Docomo taps Accenture, AWS for AI governance platform appeared first on Mobile World Live.

Anthropic raised $65 billion in its latest funding round, taking its valuation to $965 billion as investor money continues to pour into big name AI companies.

Financial Times reported Anthropic’s valuation overtook OpenAI’s following the round.

Anthropic plans to use the latest funds to advance “safety and interpretability research, expand compute to meet growing demand” for AI assistant Claude “and scale the products and partnerships our customers rely on”.

The financing was led by Altimeter Capital, Dragoneer, Greenoaks and Sequoia Capital. Other investors include private equity funds and the company’s partners. The $65 billion includes $15 billion in previously-committed cash from so-called hyperscale companies.  

Micron Technology, Samsung and SK Hynix, which Anthropic describes as “strategic infrastructure partners”, were also among the lengthy list of backers.

Cash
As with peers in the AI boom, the company is no stranger to funding rounds amounting to multiple billions of dollars.

In February, it raised $30 billion, which brought its valuation at the time to $380 billion.

Anthropic noted since that round, its Claude AI offering gained further traction with enterprises around the globe and across a range of industries, with its run-rate revenue crossing the $47 billion mark this month.

The post Anthropic approaches $1T valuation appeared first on Mobile World Live.

IoT specialist Wiliot expanded work with AT&T Business to boost its position in the physical AI field by tapping the US-based operator’s connectivity, device relationships and general expertise in data handling.

Wiliot is seeking to broaden its position in physical AI for enterprise supply chains and believes AT&T has the network connectivity, device relationships and ability in employing data to do so.

“Physical AI depends on continuous data” from the real world, VP of marketing Amir Khoshniyati said.

Deepening ties with AT&T improves Wiliot’s “ability to deploy and operate physical AI networks across large, distributed environments”.

The company is pitching a platform which handles sensing and intelligence, employing real-time data from dedicated devices. AT&T is tasked with delivering the network infrastructure, mobile connectivity and execution elements.

Certification of Wiliot gateway devices on AT&T’s network is also underway, as the company seeks more direct connectivity and standardised deployments for the enterprise space.

Wiliot stated the companies embarked on a systems integration collaboration late in 2025 covering core deployments and operational capabilities. The operator is also handling various design, installation, asset tagging and maintenance aspects.

Lee Wagner, area VP for AT&T, explained enterprises “need actionable data from the physical world” and the companies are “adding visibility at the case and asset level” to provide a fresh range of services.

Work already undertaken in retail, food and beverage, and some restaurant sectors delivered inventory accuracy improvements of 99%, Wiliot stated.

It also highlighted improvements in the time taken to ship goods to storage, a reduction in the number of staff required in receiving items and greater shipment accuracy.

The post Wiliot, AT&T make physical AI move appeared first on Mobile World Live.

Meta Platforms prepared to test paid plans for its AI services and expand availability of subscriptions for WhatsApp, Instagram and Facebook, as the technology giant looks to diversify revenue streams during a period of heavy spending.

Naomi Gleit, Meta’s head of product, explained in an Instagram video the company is “starting to roll out Facebook Plus, Instagram Plus and WhatsApp Plus with enhanced features”.

She added users accessing Meta AI will be given “more to work with, more capacity, bigger, more complex requests, and more room to create for businesses and creators”.

Bloomberg reported the social media giant will trial two consumer AI subscription tiers from next month in Singapore, Guatemala and Bolivia, while retaining a free version of the Meta AI app and website.

Meta One Plus will apparently cost $7.99 a month and target users who regularly generate AI images and videos or make heavy use of reasoning features, while Meta One Premium will be around $20 and offer the same tools but with higher usage limits.

Specific products for businesses and creators, Meta One Essential and Meta One Advanced, are also set to be offered.

WhatsApp, Instagram and Facebook offerings will be priced at around $2.99 to $3.99 a month depending on the market, Bloomberg reported. Users paying for Meta AI will gain access to those app-specific benefits.

“We’re offering premium tools that allow you to enhance presence, supercharge content, automate tasks and protect your brand” Gleit said, adding “eventually we see Meta One as the one place that brings our subscriptions together across all of our apps”.

The trials are Meta’s first attempt to charge consumers for AI features. Rivals OpenAI and Google already offer paid chatbot subscriptions.

Its move to generate subscription revenue comes during an aggressive AI investment drive.

Meta is pumping more than $10 billion into building a massive data centre campus in the US state of Indiana. Last month, the company also raised its capex forecast for 2026 to between $125 billion and $145 billion to fund AI infrastructure plans.

The post Meta adds paid tier for social media apps, eyes AI revenue appeared first on Mobile World Live.

A report by Telus Digital exposed significant security vulnerabilities across the generative AI landscape, which included finding every major model could be coaxed into unsafe behaviour under the right conditions.

In the company’s second GenAI Safety Model Benchmark, Telus Digital’s testing found some models engaged with harmful requests more than 90% of the time and stated most enterprises are dangerously underprepared to defend against them.

The testing drew on more than 620,000 adversarial tests across 34 AI models from 10 global providers: Anthropic, OpenAI, Google, Meta, Alibaba, Baidu, ByteDance, Zhipu AI, 01.AI and Mistral.

It is the most extensive AI security study Telus Digital has conducted to date, nearly doubling the scope of the first edition published in November 2025.

Attack vulnerability rates across tested models ranged from 1.3% to 93%, where a lower percentage means a safer model. 

Anthropic’s Claude models claimed five of the 10 lowest vulnerability scores, including the benchmark’s overall lowest rate, but Telus Digital noted even single-digit failure rates are unacceptable in high-stakes enterprise contexts involving money, health and reputation.

The research identified model size, reasoning capability and the creator’s overall safety approach as the strongest predictors of resilience.

Reasoning models, designed to deliberate before responding, proved significantly harder to exploit, with a 19.9% vulnerability rate compared to 55.1% for standard models which skip the reasoning step.

Smaller models are consistently the most susceptible to attacks, regardless of whether they are open-source or proprietary. The study found open source models are not inherently less safe than closed ones, with GLM 4.7 from China’s Zhipu AI outperforming many proprietary alternatives.

The benchmark also highlighted where risks cluster most sharply: privacy exploitation, fraud and cybersecurity threats remain the hardest categories for even leading models to handle.

Telus Digital also flagged a pattern called “refuse-but-engage,” where a model declines a harmful request but still provides related information which could cause harm or reputational damage.

Global AI spending is projected to reach $2.52 trillion in 2026, yet spending on AI trust, risk, and security management is projected to be just $3.4 billion, which is roughly $1 in security for every $735 spent on AI capabilities.

Meanwhile, 86% of organisations report having already experienced an AI-related security incident.

Telus Digital urges enterprises to move beyond one-time or periodic safety checks toward continuous, automated adversarial testing embedded directly into development workflows, layered with human oversight and clean data practices.

The post Telus Digital research reveals safety gaps across AI models appeared first on Mobile World Live.