Quantum leap as Caltech builds 6,100-qubit processor

A team of physicists at the California Institute of Technology has unveiled a quantum computing breakthrough, creating an array of 6,100 qubits, the largest of its kind to date.

The leap surpasses previous systems, which typically contained around a thousand qubits, and marks a step closer to practical quantum algorithms.

Researchers used caesium atoms as qubits, trapping them with laser tweezers inside an ultra-high-vacuum chamber.

These qubits maintained superposition for almost 13 seconds, nearly ten times longer than previous benchmarks. They could also be manipulated with 99.98 percent accuracy, proving that scaling up need not compromise precision.

Unlike classical bits, qubits exploit superposition, allowing a spread of probabilities instead of fixed binary states. It enables powerful computations but also demands error correction to overcome qubit fragility. The surplus qubits in this new array provide a path to large, error-corrected machines.

Physicists believe the next milestone will involve harnessing entanglement, enabling the shift from storing quantum information to processing it. If progress continues, quantum computers could soon revolutionise science by uncovering new materials, forms of matter, and fundamental laws of physics.

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

Google tests AI hosts for YouTube Music

Google is testing AI-generated hosts for YouTube Music through its new YouTube Labs programme. The AI hosts will appear while users listen to mixes and radio stations, providing commentary, fan trivia, and stories to enrich the listening experience.

The feature is designed to resemble a radio jockey but relies on AI, so there is a risk of occasional inaccuracies.

YouTube Labs, similar to Google Labs, allows the company to trial new AI features and gather user feedback before wider release. The AI hosts are currently available to a limited group of US testers, who can sign up via YouTube Labs and snooze commentary for an hour or all day.

The rollout follows Google’s Audio Overviews in NotebookLM, which turns research papers and documents into podcast-style summaries. Past AI experiments on YouTube, such as automatic dubbing, faced criticism as viewers had limited control over translations.

The AI hosts experiment shows Google’s push to integrate AI across its apps, enhancing engagement while monitoring feedback before wider rollout.

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!

Balancing chaos and precision: The paradox of AI work

In a recent blog post, Jovan Kurbalija explores why working in AI often feels like living with two competing personalities. On one side is the explorer, curious, bold, and eager to experiment with new models and frameworks. That mindset thrives on quick bursts of creativity and the thrill of discovering novel possibilities.

Yet, the same field demands the opposite. The engineer’s discipline, a relentless focus on precision, validation, and endless refinement, until AI systems are impressive and reliable.

The paradox makes the search for AI talent unusually difficult. Few individuals naturally embody both restless curiosity and meticulous perfectionism.

The challenge is amplified by AI itself, which often produces plausible but uncertain outputs, requiring both tolerance for ambiguity and an insistence on accuracy. It is a balancing act between ADHD-like energy and OCD-like rigour—traits rarely found together in one professional.

The tension is visible across disciplines. Diplomats, accustomed to working with probabilities in unpredictable contexts, approach AI differently from software developers trained in deterministic systems.

Large language models blur these worlds, demanding a blend of adaptability and engineering rigour. Recognising that no single person can embody all these traits, the solution lies in carefully designed teams that combine contrasting strengths.

Kurbalija points to Diplo’s AI apprenticeship as an example of this approach. Apprentices are exposed to both the ‘sprint’ of quickly building functional AI agents and the ‘marathon’ of refining them into robust, trustworthy systems. By embracing this duality, teams can bridge the gap between rapid innovation and reliable execution, turning AI’s inherent contradictions into a source of strength.

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

Microsoft adds AI auto-categorisation to Photos app

Microsoft is introducing an AI-powered auto-categorisation feature to the Photos app on Windows 11 for all Insider channels. The update automatically sorts images into screenshots, receipts, identity documents, and notes, making managing large photo libraries easier.

The feature is language-agnostic, recognising document types regardless of language, and aims to save time and reduce clutter.

Photos that match the AI model are grouped automatically, but users can browse categories via the app’s navigation or Search bar and manually reassign images if needed. The update adds Super Resolution to Copilot Plus PCs, enhancing low-resolution images with advanced AI.

Microsoft has included other unspecified fixes and improvements in the update, ensuring overall app performance is optimised. The company emphasises that the rollout may be gradual, so some features appear later for certain Insider users.

To access the new features, users must update the Photos app to version 2025.11090.25001.0 or higher via the Microsoft Store. The enhancements are part of Microsoft’s ongoing effort to make Windows 11 more intuitive, efficient, and AI-driven for everyday tasks.

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

UK supports JLR supply chain with £1.5 billion loan guarantee

The UK Government will guarantee a £1.5 billion loan to Jaguar Land Rover (JLR) in response to the cyber-attack that forced the carmaker to halt production.

An Export Development Guarantee, administered by UK Export Finance, will back a commercial bank loan repaid over five years to stabilise JLR’s finances and protect its supply chain.

Business Secretary Peter Kyle described the attack as a strike on the UK’s automotive sector and said the guarantee would safeguard jobs across the West Midlands, Merseyside and beyond.

Chancellor Rachel Reeves called JLR a ‘jewel in the crown’ of the UK economy, stressing that the package would protect tens of thousands of jobs directly and indirectly linked to the manufacturer.

JLR employs 34,000 people in the UK and supports an automotive supply chain of 120,000 workers, many in SMEs.

The guarantee forms part of the Government’s modern Industrial Strategy, which includes backing for electric vehicle adoption, reduced energy costs for manufacturers, and multi-billion-pound commitments to research and development.

An announcement follows ministerial visits to JLR headquarters and supplier Webasto, with ministers promising to keep working with industry leaders to get production back online and strengthen Britain’s automotive resilience.

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

OpenAI for Germany to modernise public sector with AI

SAP SE and OpenAI have announced the launch of OpenAI for Germany, a partnership to bring advanced AI solutions to the public sector.

The initiative will combine SAP’s expertise with OpenAI’s AI technology, ensuring safe, responsible use while meeting strict German data, security, and legal standards. The platform will be supported by SAP’s Delos Cloud, running on Microsoft Azure technology.

Starting in 2026, the collaboration will help public sector staff and institutions streamline tasks, automate workflows, and focus on people rather than paperwork. Customised AI will be integrated into existing systems to improve records management and data analysis.

SAP plans to expand Delos Cloud infrastructure to 4,000 GPUs to support AI workloads and will explore further investment based on demand.

OpenAI for Germany aligns with the country’s national AI strategy, which aims for AI-driven value creation of up to 10% of GDP by 2030. The ‘Made for Germany’ initiative, supported by 61 companies including SAP, has pledged over €631 billion for growth and digital modernisation.

SAP has also committed more than €20 billion to reinforce Europe’s digital sovereignty.

SAP, OpenAI, and Microsoft executives emphasised the partnership’s focus on trust, safety, and operational resilience. The initiative underscores Germany’s commitment to AI, maintaining strict standards and ensuring benefits reach all public institutions.

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

Google unveils new Gemini Robotics models

Google has unveiled two new robotics models, Gemini Robotics 1.5 and Gemini Robotics-ER 1.5, designed to help robots better perceive, plan, and act in complex environments. The models aim to enable more capable robots to complete multi-step tasks efficiently and transparently.

Gemini Robotics 1.5 converts visual information and instructions into actions, letting robots think before acting and explain their reasoning. Gemini Robotics-ER 1.5 acts as a high-level planner, reasoning about the physical world and using tools like Google Search to support decisions.

Together, the models form an ‘agentic’ framework. ER 1.5 orchestrates a robot’s activities, while Robotics 1.5 carries them out, enabling the machines to tackle semantically complex tasks. The pairing strengthens generalisation across diverse environments and longer missions.

Google said Gemini Robotics-ER 1.5 is now available to developers through the Gemini API in Google AI Studio, while Gemini Robotics 1.5 is currently open to select partners. Both models advance robots’ reasoning, spatial awareness, and multi-tasking capabilities.

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

OpenAI unveils ChatGPT Pulse for proactive updates

OpenAI has introduced a preview of ChatGPT Pulse, a feature designed to deliver proactive and personalised updates to Pro users on mobile. Instead of waiting for users to ask questions, Pulse researches chat history, feedback, and connected apps to deliver daily insights.

The updates appear as visual cards covering relevant topics, which users can scan quickly or expand for detail. Integrations with Gmail and Google Calendar are available, enabling suggestions such as drafting meeting agendas, recommending restaurants for trips, or reminding users about birthdays.

These integrations are optional and can be switched off at any time.

Pulse is built to prioritise usefulness over screen time, offering updates that expire daily unless saved or added to chat history. Early trials with students highlighted the importance of simple feedback to refine results, and users can guide what appears by curating topics or rating suggestions.

OpenAI plans to refine the feature further before expanding its availability beyond Pro users.

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

China opens digital RMB centre in Shanghai

China has launched its international operation centre for the digital RMB in Shanghai, a move seen as a significant step towards the currency’s global expansion, according to the People’s Bank of China (PBOC).

The centre will promote digital RMB internationalisation, support digital finance innovation, and enhance financial market services. It forms part of eight measures announced earlier this year by PBOC governor Pan Gongsheng at the Lujiazui Forum.

Three major platforms have been unveiled alongside the launch: a digital payment system for international use, a blockchain service platform, and a digital asset platform.

These platforms aim to broaden the application of digital technologies in finance and improve transaction efficiency.

Experts, including Tian Xuan of Tsinghua University, describe the initiative as a milestone for China’s role in global finance, strengthening its influence in shaping the future of digital payments.

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