Cloud computing

Claude Sonnet 4.5 expands developer options with rollbacks and longer-running agents

Anthropic has released Claude Sonnet 4.5, featuring a suite of new upgrades designed to enhance coding, automation, and creativity. The update enhances Claude Code, extends Computer Use, and introduces experimental tools to boost productivity and facilitate real-world applications.

Claude Code now features checkpoints, allowing developers to roll back projects to earlier versions. The Claude API has also been expanded, supporting longer-running agents to generate files such as slides, spreadsheets, and documents directly within chats.

The model’s Computer Use function has been strengthened, enabling agents to operate applications for up to 30 hours autonomously. Anthropic says Claude Sonnet 4.5 built a Slack-style app with 11,000 lines of code in one session.

A new feature, Imagine with Claude, focuses on generating creative software. The system produced a Shakespeare-themed desktop with customised scripts and performance schedules from a single prompt, highlighting its versatility.

Anthropic has maintained steady pricing for free and premium users, positioning Sonnet 4.5 as its most practical and feature-rich release yet, combining reliability with expanded creative and developer-friendly tools.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

NSW expands secure AI platform NSWEduChat across schools

Following successful school trials, the New South Wales Department of Education has confirmed the broader rollout of its in-house generative AI platform, NSWEduChat.

The tool, developed within the department’s Sydney-based cloud environment, prioritises privacy, security, and equity while tailoring content to the state’s educational context. It is aligned with the NSW AI Assessment Framework.

The trial began in 16 schools in Term 1, 2024, and then expanded to 50 schools in Term 2. Teachers reported efficiency gains, and students showed strong engagement. Access was extended to all staff in Term 4, 2024, with Years 5–12 students due to follow in Term 4, 2025.

Key features include a privacy-first design, built-in safeguards, and a student mode that encourages critical thinking by offering guided prompts rather than direct answers. Staff can switch between staff and student modes for lesson planning and preparation.

All data is stored in Australia under departmental control. NSWEduChat is free and billed as the most cost-effective AI tool for schools. Other systems are accessible but not endorsed; staff must follow safety rules, while students are limited to approved tools.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

AI-powered Opera Neon browser launches with premium subscription

After its announcement in May, Opera has started rolling out Neon, its first AI-powered browser. Unlike traditional browsers, Neon is designed for professionals who want AI to simplify complex online workflows.

The browser introduces Tasks, which act like self-contained workspaces. AI can understand context, compare sources, and operate across multiple tabs simultaneously to manage projects more efficiently.

Neon also features cards and reusable AI prompts that users can customise or download from a community store, streamlining repeated actions and tasks.

Its standout tool, Neon Do, performs real-time on-screen actions such as opening tabs, filling forms, and gathering data, while keeping everything local. Opera says no data is shared, and all information is deleted after 30 days.

Neon is available by subscription at $19.90 per month. Invitations are limited during rollout, but Opera promises broader availability soon.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Semicon Coalition unites EU on chip strategy and autonomy

European ministers have signed the Declaration of the Semicon Coalition, calling for a revised EU Chips Act 2.0 to boost semiconductor resilience, innovation, and competitiveness. The declaration outlines five priorities: collaboration, investment, skills, sustainability, and global partnerships.

The coalition, launched by the Netherlands in March, includes Austria, Belgium, Finland, France, Germany, Italy, Poland, and Spain. Other EU states joined today in Brussels, where Dutch minister Vincent Karremans presented the declaration to the European Commission.

Over fifty leading European and international semiconductor players have endorsed the declaration. This support strengthens momentum for placing end-markets at the core of the EU’s semiconductor strategy and aligns with Mario Draghi’s report on competitiveness.

The priorities include aligning EU and national funding, accelerating approvals for strategic projects, building a skilled talent pipeline, and promoting circular, energy-efficient manufacturing. International partnerships will also be deepened while safeguarding European strategic autonomy.

Minister Karremans said the strategy demonstrates Europe’s response to global tensions and its commitment to boosting semiconductor capacity, research funding, and readiness for demand in AI, automotive, energy, and defense.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Qwen3-Omni tops Hugging Face as China’s open AI challenge grows

Alibaba’s Qwen3-Omni multimodal AI system has quickly risen to the top of Hugging Face’s trending model list, challenging closed systems from OpenAI and Google. The series unifies text, image, audio, and video processing in a single model, signalling the rapid growth of Chinese open-source AI.

Qwen3-Omni-30B-A3B currently leads Hugging Face’s list, followed by the image-editing model Qwen-Image-Edit-2509. Alibaba’s cloud division describes Qwen3-Omni as the first fully integrated multimodal AI framework built for real-world applications.

Self-reported benchmarks suggest Qwen3-Omni outperforms Qwen2.5-Omni-7B, OpenAI’s GPT-4o, and Google’s Gemini-2.5-Flash, known as ‘Nano Banana’, in audio recognition, comprehension, and video understanding tasks.

Open-source dominance is growing, with Alibaba’s models taking half the top 10 spots on Hugging Face rankings. Tencent, DeepSeek, and OpenBMB filled most of the remaining positions, leaving IBM as the only Western representative.

The ATOM Project warned that US leadership in AI could erode as open models from China gain adoption. It argued that China’s approach draws businesses and researchers away from American systems, which have become increasingly closed.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Brazil to host massive AI-ready data centre by RT-One

RT-One plans to build Latin America’s largest AI data centre after securing land in Uberlândia, Minas Gerais, Brazil. The US$1.2bn project will span over one million square metres, with 300,000 m² reserved as protected green space.

The site will support high-performance computing, sovereign cloud services, and AI workloads, launching with 100MW capacity and scaling to 400MW. It will run on 100% renewable energy and utilise advanced cooling systems to minimise its environmental impact.

RT-One states that the project will prepare Brazil to compete globally, generate skilled jobs, and train new talent for the digital economy. A wide network of partners, including Hitachi, Siemens, WEG, and Schneider Electric, is collaborating on the development, aiming to ensure resilience and sustainability at scale.

The project is expected to stimulate regional growth, with jobs, training programmes, and opportunities for collaboration between academia and industry. Local officials, including the mayor of Uberlândia, attended the launch event to underline government support for the initiative.

Once complete, the Uberlândia facility will provide sovereign cloud capacity, high-density compute, and AI-ready infrastructure for Brazil and beyond. RT-One says the development will position the city as a hub for digital innovation and strengthen Latin America’s role in the global AI economy.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Gemini’s image model powers Google’s new Mixboard platform

Google has launched Mixboard, an experimental AI tool designed to help users visually explore, refine, and expand ideas both textually and visually. The Gemini 2.5 Flash model powers the platform and is now available for free in beta for users in the United States.

Mixboard provides an open canvas where users can begin with pre-built templates or custom prompts to create project boards. It can be used for tasks such as planning events, home decoration, or organising inspirational images, presenting an overall mood for a project.

Users can upload their own images or generate new ones by describing what they want to see. The tool supports iterative editing, allowing minor tweaks or combining visuals into new compositions through Google’s Nano Banana image model.

Quick actions like regenerating and others like this enable users to explore variations with a single click. The tool also allows text generation based on context from images placed on the board, helping tie visuals to written ideas.

Google says Mixboard is part of its push to make Gemini more useful for creative work. Since the launch of Nano Banana in August, the Gemini app has overtaken ChatGPT to rank first in the US App Store.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Content Signals Policy by Cloudflare lets websites signal data use preferences

Cloudflare has announced the launch of its Content Signals Policy, a new extension to robots.txt that allows websites to express their preferences for how their data is used after access. The policy is designed to help creators maintain open content while preventing misuse by data scrapers and AI trainers.

The new tool enables website owners to specify, in a machine-readable format, whether they permit search indexing, AI input, or AI model training. Operators can set each signal to ‘yes,’ ‘no,’ or leave it blank to indicate no stated preference, providing them with fine-grained control over their responses.

Cloudflare says the policy tackles the free-rider problem, where scraped content is reused without credit. With bot traffic set to surpass human traffic by 2029, it calls for clear, standard rules to protect creators and keep the web open.

Customers already using Cloudflare’s managed robots.txt will have the policy automatically applied, with a default setting that allows search but blocks AI training. Sites without a robots.txt file can opt in to publish the human-readable policy text and add their own preferences when ready.

Cloudflare emphasises that content signals are not enforcement mechanisms but a means of communicating expectations. It is releasing the policy under a CC0 licence to encourage broad adoption and is working with standards bodies to ensure the rules are recognised across the industry.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Expanded AI model support arrives in Microsoft 365 Copilot

Microsoft is expanding the AI models powering Microsoft 365 Copilot by adding Anthropic’s Claude Sonnet 4 and Claude Opus 4.1. Customers can now choose between OpenAI and Anthropic models for research, deep reasoning, and agent building across Microsoft 365 tools.

The Researcher agent can now run on Anthropic’s Claude Opus 4.1, giving users a choice of models for in-depth analysis. The Researcher draws on web sources, trusted third-party data, and internal work content—encompassing emails, chats, meetings, and files—to deliver tailored, multistep reasoning.

Claude Sonnet 4 and Opus 4.1 are also available in Copilot Studio, enabling the creation of enterprise-grade agents with flexible model selection. Users can mix Anthropic, OpenAI, and Azure Model Catalogue models to power multi-agent workflows, automate tasks, and manage agents efficiently.

Claude in Researcher is rolling out today to Microsoft 365 Copilot-licensed customers through the Frontier Program. Customers can also use Claude models in Copilot Studio to build and orchestrate agents.

Microsoft says this launch is part of its strategy to bring the best AI innovation across the industry to Copilot. More Anthropic-powered features will roll out soon, strengthening Copilot’s role as a hub for enterprise AI and workflow transformation.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

LinkedIn default AI data sharing faces Dutch privacy watchdog scrutiny

The Dutch privacy watchdog, Autoriteit Persoonsgegevens (AP), is warning LinkedIn users in the Netherlands to review their settings to prevent their data from being used for AI training.

LinkedIn plans to use names, job titles, education history, locations, skills, photos, and public posts from European users to train its systems. Private messages will not be included; however, the sharing option is enabled by default.

AP Deputy Chair Monique Verdier said the move poses significant risks. She warned that once personal data is used to train a model, it cannot be removed, and its future uses are unpredictable.

LinkedIn, headquartered in Dublin, falls under the jurisdiction of the Data Protection Commission in Ireland, which will determine whether the plan can proceed. The AP said it is working with Irish and EU counterparts and has already received complaints.

Users must opt out by 3 November if they do not wish to have their data used. They can disable the setting via the AP’s link or manually in LinkedIn under ‘settings & privacy’ → ‘data privacy’ → ‘data for improving generative AI’.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!