tag

Anthropic said Friday it has suspended access to two powerful AI models, Fable 5 and Mythos 5, to comply with a US national security order.Just three days after publicly launching Fable 5, the company said in a blog post that it received a government directive banning all foreign nationals, even ones who work at Anthropic, from accessing Fable 5 and Mythos 5 over national security concerns."The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance," it said.The company said it received the letter at 5:21 pm (2121 GMT) Friday.Axios reported that the letter came from US Commerce Secretary Howard Lutnick.The US Commerce officials did not immediately respond to a request from AFP.The firm said that the letter did not state what specifically concerned the government. However, the firm's "understanding is that the government believes it has become aware of a method of bypassing, or 'jailbreaking'" the Fable 5 model such that it could aid hacking.Fable 5, released Tuesday, is a locked-down version of Mythos 5, a cutting-edge AI model that Anthropic has held back from the public amid concerns that it had unprecedented abilities to identify software vulnerabilites — or holes in code that hackers could exploitMythos 5 — the unrestricted model — has only been released to select companies.The European Union, which gained access to Mythos earlier in June after weeks of talks, said the latest development further underlined "Europe's need for technological sovereignty"."We take note of Anthropic's statement and are assessing," said Thomas Regnier, a spokesman for the European Commission, which this month unveiled measures to slash the 27-nation bloc's dependence on America and Asia for key technologies, including AI.Anthropic said it had reviewed the "jailbreaking" method at the center of the speculation and the hacking opportunities it exposed, but it does not believe Fable 5 gives hackers capabilities that are not already available through other public models.The firm said that none of its security testers had found a "universal jailbreak," or a way to bypass it's safeguards against helping hackers."We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people," the company said."If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers."Anthropic has been locked in a legal standoff with the Trump administration for refusing to allow its technology to potentially be used for mass surveillance and autonomous weapons, leading the Pentagon to cut contracts with the company. 

Students in Qatar should build strong foundations in Arabic and their mother tongue before expanding into other languages, a Unesco education official has said, highlighting language's role in identity and learning.Speaking to Gulf Times, Farida Aboudan, head of Education Sector at Unesco's Regional Office for the Gulf States and Yemen, stressed that language should be viewed as far more than a communication tool, particularly in countries with diverse student populations such as Qatar."Language is not only a means of communication, it's not only a means for teaching and learning," she said. "It is the foundation of identity, the cultural aspects of people, the context they live in, their communities, their societies and their history."Aboudan’s comments come as Qatar continues to host learners from a wide range of linguistic and cultural backgrounds, creating a unique educational environment where local identity and global learning increasingly intersect.She noted that Unesco has worked on language and education programmes for more than 70 years and recently reinforced its position through updated guidance highlighting the importance of multilingual education.According to Aboudan, preserving Arabic and encouraging the use of learners' mother tongues should remain a priority, even as students acquire international languages such as English and French."In Qatar we have learners from different backgrounds and it's important to preserve and encourage the use of Arabic and the use of mother tongue more generally," she said. "At the same time, other languages will help students in their careers and professional progression."She emphasised that strong literacy and language skills in a student's first language can support wider educational achievement."The principle is that we need to start with the mother tongue," Aboudan explained. "We need to ensure that learners have a strong foundation and strong skills in their mother tongue and then we can move ahead with more success in other domains of education as well."Looking to the future, she pointed to what she described as a growing localisation trend across Gulf education systems.Traditionally, Aboudan said, national schools have primarily used Arabic while international schools have relied on English, French or other dominant languages.However, she believes the region is increasingly moving towards educational models that blend international standards with local culture, values and languages."We combine global standards with local knowledge, local values and local languages," she said. "We need to start from what learners already know, from their families, communities and places."Aboudan also highlighted the emergence of place-based education, an approach that connects learning to students' local environments and experiences.She described the trend as a positive development for the Gulf region, saying that it can help students remain connected to their identity while preparing them for an increasingly globalised world. 

In the race to build larger AI models, one weakness remains stubbornly unresolved: memory. Dubai & London-based SeKondBrain is betting that the next frontier in AI will be systems that preserve context, meaning and knowledge over time. That matters because most skilled work is cumulative. An electrician diagnoses a fault faster because she has seen a similar pattern before. A writer improves through successive drafts. A chef sharpens instinct through repetition. Knowledge has value on its own, but real expertise comes from what is retained, refined and carried forward. Today’s AI systems are still poor at that kind of continuity. Many leading systems try to compensate by processing ever-larger volumes of raw text, known as tokens, in pursuit of better outputs. By expanding their context windows, these models can read more, summarize more and generate more than ever before. But when it comes to preserving understanding over time, they still falter. McKinsey estimates that employees spend one-fifth of their time searching for, interpreting, or recreating information. Meanwhile, research cited by Stanford’s Institute for Human-Centered AI suggests that as context windows expand, consistency and reasoning can degrade. AI was supposed to reduce these frictions, not reinforce them. Yet in many cases, it still accelerates fragmented work rather than fixing it. SeKondBrain AI is tackling this problem directly, with a focus on preserving meaning and memory. “Tokens are sand. Concepts are Lego,” says Duggal, by which he means that without conceptual structure, raw information slips through our fingers. That thesis places SeKondBrain within a broader shift toward neurosymbolic AI. While purely generative systems are powerful at producing text, they often struggle with explainability, structured reasoning, and durable memory. Neurosymbolic systems address that gap by combining neural pattern recognition with symbolic reasoning. The field is rapidly gaining momentum, with the global neurosymbolic AI market projected to reach $4.87 billion by 2034 and foundation capital pitting the context layer as a 1 trillion-dollar market. Founded less than a year ago by entrepreneur Sachin Duggal, SeKondBrain is building what it calls a “cognitive operating system” – a shared intelligence layer designed to sit beneath people, enterprise software, and AI agents. Modern AI systems, in Duggal’s view, are increasingly powerful engines with an obvious missing layer – “autonomous cars without a map.” The car may be fast and its sensors impressive. Butwithout a stable map of the world – what matters, what’s changing, what’s allowed, and so on – it’s still can’t get out of the driveway. In contrast, SeKondBrain’s platform expresses itself through interconnected surfaces. Kora functions as a conversational intelligence layer, preserving context across conversations, teams, and calendars. Kanvas serves as a workspace where documents, ideas, and digital artifacts are connected by concept rather than location. And the Kemory allows AI to have a shared contextual memory so you don’t need to copy and paste between your chats. For businesses there is also KompanyBrain; a single pane of glass to bring all your organisational & customer intelligence and knowledge together with early design customers in Consultative Industries like Legal and Consulting. Early test cases were intentionally practical. One is an “Experience Finder” – a system that can answer questions like, “Who inside the organisation has solved this problem before?” or “Which team has worked on this client, sector or regulatory issue?” and the other was for law firms to simply have a single pane of glass across client workflow, files and communications. SeKondBrain sees these systems evolving into organisation-specific knowledge layers for sectors including law, consulting, finance and government, where expertise is often buried across documents, meetings, emails and internal conversations. This would also support faster software development, since new tools no longer need to worry about the messy data layer causing information dementia in modern day companies. That ambition also helps explain why Duggal is building from Dubai. A system designed to connect fragmented knowledge across teams, sectors and markets needs more than technical talent. It needs proximity to global businesses, government buyers, capital, and commercial corridors where ideas can move quickly from pilot to deployment. In one sense, the choice extends Dubai’s historic role as an entrepôt: a place where goods, capital and talent move through, are refined, and then exported outward. SeKondBrain applies that logic to knowledge infrastructure. Built from the Dubai corridor to London, refined with a team that convenes in the emirate each month, and designed to ship globally, it treats Dubai less as a market to sell into than as a platform from which to build. But it also reflects a deeper shift. For decades, advanced technology was assumed to be built elsewhere – in Silicon Valley, London or Beijing – and then sold into the Gulf. That assumption is now being challenged. The UAE is no longer just a destination market for global technology; it is increasingly a place where globally relevant companies are built. The economic momentum is significant. The UAE’s AI market is projected to reach $46 billion by 2030, with AI expected to contribute close to 14 percent of national GDP. But beyond the economics lies a more consequential question: who will control the underlying architecture of intelligence itself? The last era of the internet was shaped by those who owned the servers, platforms and cloud infrastructure. The next may be shaped by those who own the systems that preserve memory, meaning and shared cognition. SeKondBrain’s wager is that part of that foundational layer can be built in Arabic and English, developed from Dubai, and exported to the world. It’s a bold bet. But in a crowded AI market, the ability to make intelligence persist – to remember, adapt and compound over time – may prove to be one of the most valuable opportunities of all; because as Duggal says “models commodities, knowledge compounds” 

As China prepares for Lunar New Year holidays starting on Sunday, rivals to DeepSeek are scrambling to release artificial-intelligence models a year after it burst onto the scene with its game-changing R1 and V3 models.With DeepSeek set to launch its next-generation V4 ‌model soon, according to tech news site The Information, many other Chinese AI firms have released ​or are preparing to launch their own ‌models in the hopes of stealing the spotlight — or at least avoid being off-guard again ‌during this year's ⁠Spring Festival.Below are the companies ‌and models looking to make a Spring Festival ‌splash. DEEPSEEKThe Hangzhou-based startup's V4 would replace last year's V3 model, which powered the AI assistant app that overtook ChatGPT to ⁠become the top-rated free application available on Apple's App Store in the US Investors and industry insiders are also on the lookout for R2, successor to the R1 model.This week DeepSeek fuelled anticipation when its web and mobile chatbot upgraded its "context window" — the amount of information it can remember and handle in a single task, from 128,000 to1 mn tokens, the unit of data processed by the AI model.This means the chatbot can now process book-length passages of text to answer a single user command. BYTEDANCEThe makers of short-video ​platform TikTok released the Doubao 2.0 chatbot, the company said on Saturday. The new model is capable of deep thinking and long, multi-step task execution that matches the capabilities of US rival OpenAI's GPT 5.2 and Google's Gemini 3 Pro, Bytedance said.Doubao ‌is China's most popular AI chatbot app, ⁠in terms of weekly active ​users, according to data published by Quest Mobile in December.Thursday's release of video-generation AI model ​Seedance 2.0 has generated comparisons to DeepSeek's global rise, going viral on Chinese social media and drawing widespread praise on X, including from the platform's owner, Elon Musk. Seedance 2.0 can produce high-quality cinematic videos based on a few prompts, or even one.The tech giant released picture-generation model Seedream 5.0 Lite on Friday. ALIBABAAlibaba, the first Chinese firm to respond to DeepSeek's viral ascent last year, with Qwen 2.5-Max, is preparing to launch Qwen 3.5.The e-commerce giant's Qwen app is riding a wave of growing domestic usage after it spent 3bn yuan ($400mn) last week on a coupon giveaway campaign to promote "agentic commerce", where AI handles consumers' online shopping.This drove more than 120mn consumer orders in the six days through Wednesday, the company said. ZHIPUZhipu ‌AI released its open-source GLM-5 model on ‌Wednesday, with enhanced coding capabilities and the ability ⁠to perform long-running agent tasks.Zhipu is considered one of China's "AI tigers" — promising startups vying with the US to win ⁠the AI race. Zhipu went public on the ⁠Hong Kong Stock Exchange last month, alongside rival MiniMax, another AI tiger.Both stocks have rallied strongly as investors bet on the companies benefiting from China's AI boom. Zhipu plans a secondary listing in Shanghai, a regulatory filing on Friday showed. MINIMAXMiniMax released its M2.5 open-source model on its overseas agent website on Wednesday. The company's Hong Kong listing raised HK$4.8bn ($620mn), higher than Zhipu's $558mn.Shanghai-based MiniMax has developed popular apps like Hailuo AI, a video generation tool, and Talkie, a ​character interaction app that enables users to engage with AI-powered virtual personas. TENCENTTencent's Hunyuan team on Tuesday released a low-storage, compressed AI model, HY-1.8B-2Bit, designed to be used on consumer hardware including mobile phones. IFLYTEKiFlytek on Wednesday released Spark X2, trained entirely on Chinese-made chips. The company said the upgrade focuses on practical deployment in sectors including education, healthcare, automotive and agent-based applications. NETEASE YOUDAONetEase Youdao on Wednesday launched LobsterAI, a desktop-level personal assistant agent that can perform tasks such as information retrieval, scheduling and data analysis by executing workflows locally on a user's computer after authorisation.The product supports mobile and PC connections and allows remote interaction via enterprise apps popular among Chinese companies such as DingTalk and Feishu. DEXMALEmbodied-intelligence startup ‌Dexmal on Tuesday unveiled DM0, ​an AI model designed for robot-related scenarios. DM0 integrates multimodal internet data with driving, navigation and robotic operation data, and was trained across multiple robot platforms. 

The development of increasingly powerful models is central to the unfolding AI revolution. But this revolution has a second, equally important component: the adaptation and adoption of AI models across the economy, both to lower the cost of existing products and services, and to create new or improved products and services capable of advancing economic and social development. Whereas model development is happening largely in the United States and China, diffusion can and must take place everywhere.Overall, AI will follow a J-curve pattern. At first, there is a huge amount of investment – in areas like physical infrastructure, software, business-model adaptation, data consolidation, and human-capital development – which does not yield immediate benefits. During this period, there is downward pressure on productivity, broadly defined to include benefits not measured by conventional national income accounts.Then the technology’s value-creation potential kicks in, and the curve slopes upward. Since we haven’t yet reached this point, it is impossible to say exactly what this upswing will look like – the J-curve’s height and slope. By and large, investors seem to be betting on a massive payoff, but a distinct sense of uncertainty still permeates discussions about AI, and some predict that the technology will fall short of expectations, leading to a bust. Who turns out to be right will depend far more on diffusion than development.So far, AI diffusion has been uneven, with some sectors (especially technology, finance, and professional services) embracing the technology, and others (including large-employment sectors like health care and construction) lagging behind. While such disparities are not surprising at this point, their persistence would lead to a flatter J-curve, representing muted returns on today’s investments and delays in growth and productivity gains. Put differently, whether or not we currently have an AI investment bubble will be largely determined by the pattern and speed of diffusion in the next few years.Diffusion happens through multiple channels, the fastest of which is arguably software-as-a-service (SaaS) providers. Providers such as Google Search, Microsoft Office, Copilot by Notion, Salesforce, and Adobe are already embedding AI into their offerings. AI can also be incorporated into scientific processes relatively quickly. And with the major developers of large language and multimodal models providing application programming interfaces (APIs) that allow for the quick creation of tailored AI models, progress may pick up in other areas.Open-source models – so far seen more in China than in the US – create even more opportunity, because they enable increased specialization and competition, including from smaller firms and countries that lack the massive computing infrastructure needed for the largest models. But there are still barriers to entry: a reliable electricity supply, robust computing capacity, and accessible mobile-internet connectivity are prerequisites to broad adoption.Trade – especially of inputs like advanced semiconductors – also makes a difference. So does human capital: from advanced AI engineering and high-level strategic management to user-related skills, an economy needs to ensure access to an array of capabilities through education, reskilling, and labor mobility. The final piece of the puzzle is data. Where data systems are fragmented, incomplete, inaccurate, or inaccessible, training effective models will be slow, at best.While AI diffusion depends significantly on private-sector initiatives, policy frameworks and regulatory structures also matter. China’s leaders understand this. As Huawei founder Ren Zhengfei recently observed, China has adopted a practical approach aimed at using AI to address real-world development and economic challenges. So, while developing increasingly capable large models is a high priority, so is deploying AI broadly, in order to secure the rapid gains in service quality, efficiency, and productivity that will be needed to offset the effects of rapid population aging.China’s government is actively directing innovators toward these outcomes. Beyond encouraging the large tech platforms to build open-source models, China’s government has tasked them with developing or enabling applications in specific sectors, such as autonomous driving, health care, robotics (in manufacturing and logistics), supply-chain management, and green technologies. China’s government also regularly sponsors developer conferences and competitions.Such efforts have paid off. For example, China accounts for over 30% of total global manufacturing output. In 2024, China accounted for 54% of all robot installations globally. The country now boasts almost half the world’s installed robots – at just over 2mn. Relative to the US, China’s policy framework is much more engaged and geared toward providing direction with respect to applications and adoption across sectors in the economy. By contrast, US tech giants and well-funded AI startups are pushing the boundaries of large models, often in pursuit of artificial general intelligence and artificial superintelligence. While diffusion channels are open, their use is being left largely up to the private sector.That may work in a few sectors, like tech, finance, and professional services, with the resources and know-how to experiment and then adopt. But private actors alone are unlikely to address the factors inhibiting AI adoption in specific sectors, such as data fragmentation, capacity deficiencies, regulatory hurdles, and scale problems. The likely – and unnecessary – result is a two-speed pattern of diffusion, leading to subpar economic growth, negative distributional outcomes, and the erosion of the economic underpinnings of national security.When it comes to defence, the US government has long recognized that some state guidance is appropriate to ensure that private-sector innovation advances public goals. AI diffusion demands a similar approach. Something like this hybrid, active, pragmatic, and sector-specific approach is needed across a wide swath of the economy. Failure to do so will result in subpar economic growth, problematic distributional outcomes, and a weakening of the economic underpinnings of national security.When it comes to diffusion, watching, waiting, and hoping is not a strategy. —Project Syndicate(Michael Spence, a Nobel laureate in economics, is Emeritus Professor of Economics and a former dean of the Graduate School of Business at Stanford University and a co-author (with Mohamed A El-Erian, Gordon Brown, and Reid Lidow) of Permacrisis: A Plan to Fix a Fractured World (Simon & Schuster, 2023)). 

China’s dominance in clean energy has put the country on a singularly strong footing when it comes to competing with the rest of the world — particularly the US — in building artificial intelligence.That’s according to Julian Wentzel, chief sustainability officer at HSBC Holdings, who says an economy built on renewable energy brings with it advantages that can’t be replicated by fossil fuels.“China has put themselves in a very unique position in terms of the energy requirement to fuel their economy and ultimately their AI architecture,” Wentzel said in an interview.The vast build-out of clean energy in China — the country is on track to once again break its own record in installing renewable power this year — “enhances their cost of capital,” he said.Once renewable energy infrastructure is built and the upfront investment has been paid off, producing extra energy carries effectively no incremental cost; fossil fuels, in contrast, require ongoing costs for extraction, transport, refining and distribution, he said.“Once you’ve got the architecture in place, as the demand grows, you can deliver that demand at zero cost,” Wentzel said. As the “percentage cost of every incremental kilojoule of power relative to total GDP declines over time,” it becomes “a very powerful lever to the underlying growth of an economy.”The comments stand in contrast to the policy position of the government in the US, where Energy Secretary Chris Wright has argued that a rapid transition to clean energy will raise energy costs and hurt economic growth. And for now, fossil fuels continue to provide a major share of the energy powering AI data centres.Microsoft Corp Chief Executive Officer Satya Nadella said recently the supply of power, rather than the availability of semiconductors, accounted for the biggest bottleneck in data centre capacity. And by some estimates, the energy needs of existing and planned AI infrastructure in the US can’t be met with current supply.That dynamic has created an opportunity for oil majors to cash in on the enormous demand for energy that will be needed to power data centres. Chevron Corp said on Wednesday it will provide natural gas-fired power to a data centre in West Texas, the beginning of a new line of business for the company to capitalise on the AI boom.The global race to dominate AI depends on an array of factors that includes chips and supply chains as well as rare earths and key metals such as copper. But energy supply is key, and because renewables are low-cost to run once infrastructure is built, countries that have greater access to them have an advantage, Wentzel said.China is challenging developments in the West not just due to its dominance in cheap renewable energy, but also due to its approach to building artificial intelligence. That became apparent earlier this year, when startup DeepSeek indicated the country is capable of producing AI at a much lower cost and greater energy efficiency than US rivals.China’s growing dominance is also shaping talks at the COP30 summit in Belem, Brazil, where California Governor Gavin Newsom took several opportunities to warn that the US risks losing out on numerous fronts.One of the great abdications of the climate fight is “the own goal of the president of the US who simply doesn’t understand how enthusiastic President Xi is that the Trump administration is nowhere at COP30,” Newsom said.The US and legacy automakers “better wake up to that,” Newsom said at a press conference. “This is about economic power.”China manufactures about 80% of the world’s solar panels, supplies some 60% of the planet’s wind turbines, 70% of its electric vehicles and 75% of batteries, all at a lower financial cost than the West.To be sure, though China is adding unprecedented amounts of wind and solar, it’s still investing heavily in fossil fuels. That includes coal, which is one of the reasons the country produces almost 30% of global emissions.Wentzel said economies that rely more on clean energy are also more likely to reduce volatility in inflation.“Removing dependence on fuel commodities reduces capital account fluctuations, exposure to inflation and price swings,” he said. “As renewable systems scale, energy costs as a share of GDP can fall, strengthening the financial efficiency of the system and supporting higher economic growth.”

The Institute of Internal Auditors (IIA) held a workshop on ‘Using Generative AI for Internal Auditors’, bringing together 124 professionals from across industries to explore how Artificial Intelligence is transforming the future of auditing. The workshop offered practical insights into leveraging Generative AI tools to enhance audit procedures, enabling auditors to perform their roles more efficiently and add greater value to their organisations. Participants also gained hands-on experience with AI tools and engaged in dynamic discussions on integrating technology into everyday audit practices. The event was led by trainers Hatem el-Safty and Yousif Hussain, who shared their deep expertise on AI adoption and its applications in the internal audit profession. During the workshop, they emphasised the transformative potential of Generative AI in the internal audit profession. They highlighted how AI can assist auditors in performing audit procedures more efficiently. The trainers also emphasised that while AI can significantly enhance audit quality and speed, the human auditor’s judgment, ethics, and professional scepticism remain crucial. The experts also encouraged participants to embrace technology as a collaborative tool rather than a replacement, underscoring the need for auditors to continuously upskill and adapt in an increasingly digital audit environment.

The Ministry of Municipality has announced the launch of the pilot version of the Institutional Excellence and Knowledge Transfer System, based on artificial intelligence (AI) technologies. This is the first step of its kind in the region, coinciding with the launch of the evaluation process for the second edition of the Qatar Government Excellence Award.The new system was launched in collaboration between the Information Systems Department and the Planning, Quality, and Innovation Department. It represents a qualitative shift in the concept of institutional excellence, leveraging artificial intelligence tools to achieve superior performance and ensure the sustainability of institutional excellence through continuous improvement, accuracy, and speed of completion.The ministry explained in a statement that the system aims to enhance awareness of the concepts of excellence and innovation among leaders and decision-makers, and to spread a culture of institutional excellence within the ministry, keeping pace with the pace of digital transformation and modern institutional development trends. Excellence has become a strategic necessity dictated by global competition. The new system’s most prominent features include its role as a virtual assistant for standards leaders, rapid access to results and outputs, enhanced awareness and dissemination of a culture of institutional excellence, integrated connectivity between relevant standards and practices, the ability to extract evidence and information quickly and accurately, and the provision of an innovative, interactive environment to enhance employee engagement.The ministry emphasised that the system represents more than just a technical tool; it is a strategic partner in the institutional excellence journey and a driver of knowledge innovation. It noted that it was developed through extensive collaborative efforts and working sessions with partners from AI technology providers, establishing it as a pioneering model for digital transformation in the field of institutional excellence and innovative work.

The Public Works Authority (Ashghal) always encourages the use of technological advancements to serve the public, a senior official has said.“In road projects, we have intelligent transportation systems. We have autonomous vehicles. We even use artificial intelligence (AI). We use laser scanning technology for road networks to map any potential impacts on the roads,” said Engineer Abdulrahman Ahmed al-Ansari, Senior Projects Engineer at the Office of President of Ashghal, while talking to Qatar TV.He added that new technologies change the level of services in the sewage network.“We always encourage the use of the latest technologies in cooperation with our partners. There are the latest technologies for roads. There are various and modern technologies in sewage networks. Some of these include robots and drones,” he said.Regarding the projects that were recently awarded by Ashghal, al-Ansari said, these contracts are part of Ashghal’s strategic plan to enhance infrastructure. “These contracts include roads, public buildings, and drainage networks. Their goal is to improve the daily quality of life for the public, both citizens and residents. Through this role, we contribute to Qatar National Vision 2030 in terms of sustainability and improving services,” he said.Ashghal recently announced 13 new contracts worth QR 12bn to enhance the infrastructure of road and drainage networks and public buildings and improve the quality of life in Qatar. The new projects that were awarded include road network operation and maintenance works, drainage networks operation and maintenance works, construction of three new schools, improving safety and fire systems in 40 existing schools, refurbishment of the Psychiatric Hospital of Hamad Medical Corporation, and renovation of Al Zubara Horse Breeding Farm.

Japan's Nikkei share average closed at a record high on Wednesday, fighting back from early losses on optimism buoyed by investments in artificial intelligence. The Nikkei 225 Index rose 0.3% to close at an unprecedented 45,630.31, recovering from earlier session losses as wide as 0.6%. The broader Topix gauge gained 0.2%. The Nikkei's largest gainers were industrial manufacturer IHI, which rose 9.7%, followed by SoftBank Group, which jumped 6%. The sharpest falls were in shares of utility Tokyo Electric Power, which was down 4.89%, followed by video game maker Nexon, which sank 3.8%.

NEXX, a logistics AI (artificial intelligence) company, in association with Zipto Supply Chain, a leading Chinese cross-border E-commerce logistics provider, is expanding into Qatar market as it establishes advanced smart fulfillment center at Milaha Logistics City, Qatar, to enhance cross-border E-commerce logistics capabilities in the region.In this regard, NEXX officially announced strategic partnerships with Zipto Supply Chain and Middle East delivery leader iMile, during the Belt and Road Summit held in Hong Kong."Together with Zipto's expertise in Chinese market access and iMile's last-mile excellence, powered by our AI-driven fulfillment center, we are positioned to transform the region's logistics landscape and revolutionise service standards in this sector," said Hui Ka, Oscar, chief executive officer of NEXX.Operated jointly by NEXX, Milaha and Hong Kong E-commerce logistics company KEC, the 5,000sqm smart fulfillment center is equipped with an agentic AI management system, automated sorting robots, and pharmaceutical logistics certification.It offers end-to-end warehousing and fulfillment services tailored for cross-border B2C E-commerce customers. The center also supports B2B operations and features a bonded warehouse. It is scheduled to commence full operations in the fourth quarter of this year.On NEXX's strategic partnership with Zipto to expand into the Qatar market, this partnership will see Zipto utilise the former's advanced smart fulfillment center as its primary Qatar operational base, harnessing the facility's sophisticated automation capabilities to serve Chinese E-commerce businesses expanding into the Qatari market, with planned subsequent expansion into the UAE.In a complementary agreement, NEXX has partnered with iMile, which will establish its Qatar headquarters within NEXX's smart fulfillment center, utilising the facility's intelligent logistics infrastructure to enhance and expand its delivery services across the country through integrated technological solutions."We are pleased to support NEXX and its partners Zipto and iMile as they bring innovative logistics solutions to Qatar. Our commitment to fostering international collaboration and sustainable business growth is strengthened by these important partnerships, which will position Qatar as a central player in the region's E-commerce landscape," said Sheikh Ali Alwaleed al-Thani, chief executive officer of Invest Qatar.NEXX had recently announced a strategic investment from Rasmal Ventures — the first independent venture capital fund supported by the Qatar Investment Authority (QIA). It disclosed that Ibrahim al-Derbasti, executive vice president of Offshore and Marine at Milaha, as co-founder of NEXX Middle East.

The Qatar Investment Authority (QIA) has invested in Anthropic, a leading AI safety and research company, and creators of the Claude family of AI models.Anthropic’s Series F fundraise of $13bn, led by ICONIQ, brings Anthropic’s valuation at $183bn post-money and was co-led by Fidelity Management & Research Company and Lightspeed Venture Partners. The investment reflects Anthropic’s unprecedented velocity and reinforces its position as the leading intelligence platform for enterprises, developers, and power users.The QIA joins significant investors in this round including Altimeter, Baillie Gifford, BlackRock, Blackstone, Coatue, D1 Capital, General Atlantic, General Catalyst, GIC, Growth Equity at Goldman Sachs Alternatives, Insight Partners, Jane Street, Ontario Teachers’ Pension Plan, T. Rowe Price Associates, Inc and T. Rowe Price Investment Management, Inc, TPG, WCM Investment Management, and XN.Chief Financial Officer of Anthropic Krishna Rao said: “From Fortune 500 companies to AI-native startups, our customers rely on Anthropic’s frontier models and platform products for their most important, mission-critical work.” “We are seeing exponential growth in demand across our entire customer base. This financing demonstrates investors’ extraordinary confidence in our financial performance and the strength of their collaboration with us to continue fuelling our unprecedented growth.”The Series F investment will expand Anthropic’s capacity to meet growing enterprise demand, deepen the company’s safety research, and support international expansion as Anthropic continues building reliable, interpretable, and steerable AI systems.