Cloud computing

AI Scientist Kosmos links every conclusion to code and citations

OpenAI chief Sam Altman has praised Future House’s new AI Scientist, Kosmos, calling it an exciting step toward automated discovery. The platform upgrades the earlier Robin system and is now operated by Edison Scientific, which plans a commercial tier alongside free access for academics.

Kosmos addresses a key limitation in traditional models: the inability to track long reasoning chains while processing scientific literature at scale. It uses structured world models to stay focused on a single research goal across tens of millions of tokens and hundreds of agent runs.

A single Kosmos run can analyse around 1,500 papers and more than 40,000 lines of code, with early users estimating that this replaces roughly six months of human work. Internal tests found that almost 80 per cent of its conclusions were correct.

Future House reported seven discoveries made during testing, including three that matched known results and four new hypotheses spanning genetics, ageing, and disease. Edison says several are now being validated in wet lab studies, reinforcing the system’s scientific utility.

Kosmos emphasises traceability, linking every conclusion to specific code or source passages to avoid black-box outputs. It is priced at $200 per run, with early pricing guarantees and free credits for academics, though multiple runs may still be required for complex questions.

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

NVIDIA brings RDMA acceleration to S3 object storage for AI workloads

AI workloads are driving unprecedented data growth, with enterprises projected to generate almost 400 zettabytes annually by 2028. NVIDIA says traditional storage models cannot match the speed and scale needed for modern training and inference systems.

The company is promoting RDMA for S3-compatible storage, which accelerates object data transfers by bypassing host CPUs and removing bottlenecks associated with TCP networking. The approach promises higher throughput per terabyte and reduced latency across AI factories and cloud deployments.

Key benefits include lower storage costs, workload portability across environments and faster access for training, inference and vector database workloads. NVIDIA says freeing CPU resources also improves overall GPU utilisation and project efficiency.

RDMA client libraries run directly on GPU compute nodes, enabling faster object retrieval during training. While initially optimised for NVIDIA hardware, the architecture is open and can be extended by other vendors and users seeking higher storage performance.

Cloudian, Dell and HPE are integrating the technology into products such as HyperStore, ObjectScale and Alletra Storage MP X10000. NVIDIA is working with partners to standardise the approach, arguing that accelerated object storage is now essential for large-scale AI systems.

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

NotebookLM gains automated Deep Research tool and wider file support

Google is expanding NotebookLM with Deep Research, a tool designed to handle complex online inquiries and produce structured, source-grounded reports. The feature acts like a dedicated researcher, planning its own process and gathering material across the web.

Users can enter a question, choose a research style, and let Deep Research browse relevant sites before generating a detailed briefing. The tool runs in the background, allowing additional sources to be added without disrupting the workflow or leaving the notebook.

NotebookLM now supports more file types, including Google Sheets, Drive URLs, PDFs stored in Drive, and Microsoft Word documents. Google says this enables tasks such as summarising spreadsheets and quickly importing multiple Drive files for analysis.

The update continues the service’s gradual expansion since its late-2023 launch, which has brought features such as Video Overviews for turning dense materials into visual explainers. These follow earlier additions, such as Audio Overviews, which create podcast-style summaries of shared documents.

Google also released NotebookLM apps for Android and iOS earlier this year, extending access beyond desktop. The company says the latest enhancements should reach all users within a week.

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

Qwen relaunch aims to unify Alibaba’s mobile AI ecosystem

Alibaba is preparing a major overhaul of its mobile AI apps, renaming Tongyi as Qwen and adding early agentic features. The update aims to make Qwen resemble leading chatbots while linking AI tools to Taobao and other services. Alibaba also plans a global version once the new design stabilises.

Over one hundred developers are working on the project as part of wider AI investments. Alibaba hopes Qwen can anchor its consumer AI strategy and regain momentum in a crowded market. It still trails Doubao and Yuanbao in user popularity and needs a clearer consumer path.

Monetisation remains difficult in China because consumers rarely pay for digital services. Alibaba thinks shopping features will boost adoption by linking AI directly to e-commerce use. Qwen will stay free for now, allowing the company to scale its user base before adding paid options.

Alibaba wants to streamline its overlapping apps by directing users to one unified Qwen interface. Consolidation is meant to strengthen brand visibility and remove confusion around different versions. A single app could help Alibaba stand out as Chinese firms race to deploy agentic AI.

Chinese and US companies continue to expand spending on frontier AI models, cloud infrastructure, and agent tools. Alibaba reported strong cloud growth and rising demand for AI products in its latest quarter. The Qwen relaunch is its largest attempt to turn technical progress into a viable consumer business.

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

Google launches Private AI Compute for secure cloud-AI

In a move that underscores the evolving balance between capability and privacy in AI, Google today introduced Private AI Compute. This new cloud-based processing platform supports its most advanced models, such as those in the Gemini family, while maintaining what it describes as on-device-level data security.

The blog post explains that many emerging AI tasks now exceed the capabilities of on-device hardware alone. To solve this, Google built Private AI Compute to offload heavy computation to its cloud, powered by custom Tensor Processing Units (TPUs) and wrapped in a fortified enclave environment called Titanium Intelligence Enclaves (TIE).

The system uses remote attestation, encryption and IP-blinding relays to ensure user data remains private and inaccessible; ot even Google’s supposed to gain access.

Google identifies initial use-cases in its Pixel devices: features such as Magic Cue and Recorder will benefit from the extra compute, enabling more timely suggestions, multilingual summarisation and advanced context-aware assistance.

At the same time, the company says this platform ‘opens up a new set of possibilities for helpful AI experiences’ that go beyond what on-device AI alone can fully achieve.

This announcement is significant from both a digital policy and platform economy perspective. It illustrates how major technology firms are reconciling user privacy demands with the computational intensity of next-generation AI.

For organisations and governments focused on AI governance and digital diplomacy, the move raises questions about data sovereignty, transparency of remote enclaves and the true nature of ‘secure ‘cloud processing.

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

Microsoft brings smarter search to Copilot

Microsoft is expanding Copilot with more precise citations that link directly to publisher sources. Users can also open aggregated references for each answer to review context. The emphasis sits on trust, control, and transparent sourcing throughout the experience.

A new dedicated search mode within Copilot delivers more detailed results when queries require specific information.

Summaries appear alongside links, enabling users to verify evidence and make informed decisions quickly. Industry coverage highlights the stronger focus on verifiable sources and publisher visibility.

The right pane offers a ‘Show all’ list of sources used in responses. Source-based citation pills replace opaque markers to aid credibility checks and exploration. Design choices aim to empower people to stay in control while navigating complex topics.

Updates are live across copilot.com, mobile apps, and Copilot in Edge, with more refinements expected. Microsoft positions the changes within a human-centred strategy where AI supports curiosity safely. Broader Copilot enhancements across Windows and Edge continue in parallel roadmaps.

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

Private AI Compute by Google blends cloud power with on-device privacy

Google introduced Private AI Compute, a cloud platform that combines the power of Gemini with on-device privacy. It delivers faster AI while ensuring that personal data remains private and inaccessible, even to Google. The system builds on Google’s privacy-enhancing innovations across AI experiences.

As AI becomes more anticipatory, Private AI Compute enables advanced reasoning that exceeds the limits of local devices. It runs on Google’s custom TPUs and Titanium Intelligence Enclaves, securely powering Gemini models in the cloud. The design keeps all user data isolated and encrypted.

Encrypted attestation links a user’s device to sealed processing environments, allowing only the user to access the data. Features like Magic Cue and Recorder on Pixel now perform smarter, multilingual actions privately. Google says this extends on-device protection principles into secure cloud operations.

The platform’s multi-layered safeguards follow Google’s Secure AI Framework and Privacy Principles. Private AI Compute enables enterprises and consumers to utilise Gemini models without exposing sensitive inputs. It reinforces Google’s vision for privacy-centric infrastructure in cloud-enabled AI.

By merging local and cloud intelligence, Google says Private AI Compute opens new paths for private, personalised AI. It will guide the next wave of Gemini capabilities while maintaining transparency and safety. The company positions it as a cornerstone of responsible AI innovation.

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

AI-powered Google Photos features land on iOS, search expands to 100+ countries

Google Photos is introducing prompt-based edits, an ‘Ask’ button, and style templates across iOS and Android. In the US, iPhone users can describe edits by voice or text, with a redesigned editor for faster controls. The rollout builds on the August Pixel 10’s debut of prompt editing.

Personalised edits now recognise people from face groups, so you can issue multi-person requests, such as removing sunglasses or opening eyes. Find it under ‘Help me edit’, where changes apply to each named person. It’s designed for faster, more granular everyday fixes.

A new Ask button serves as a hub for AI requests, from questions about a photo to suggested edits and related moments. The interface surfaces chips that hint at actions users can take. The Ask experience is rolling out in the US on both iOS and Android.

Google is also adding AI templates that turn a single photo into set formats, such as retro portraits or comic-style panels. The company states that its Nano Banana model powers these creative styles and that templates will be available next week under the Create tab on Android in the US and India.

AI search in Google Photos, first launched in the US, is expanding to over 100 countries with support for 17 languages. Markets include Argentina, Australia, Brazil, India, Japan, Mexico, Singapore, and South Africa. Google says this brings natural-language photo search to a far greater number of users.

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

€5.5bn Google plan expands German data centres, carbon-free power and skills programmes

Google will invest €5.5bn in Germany from 2026 to 2029, adding a Dietzenbach data centre and expanding its Hanau facility. It will expand offices in Berlin, Frankfurt, and Munich, and launch skilling and a first German heat-recovery project. Estimated impact: ~€1.016bn GDP and ~9,000 jobs annually.

Dietzenbach will strengthen German cloud regions within Google’s 42-region network, used by firms such as Mercedes-Benz. Google Cloud highlights Vertex AI, Gemini, and sovereign options for local compliance. Continued Hanau investment supports low-latency AI workloads.

Google and Engie will extend 24/7 Carbon-Free Energy in Germany through 2030, adding new wind and solar. The portfolio will be optimised with storage and Ørsted’s Borkum Riffgrund 3. Operations are projected to be 85% carbon-free in 2026.

A partnership with Energieversorgung Offenbach will utilise excess data centre heat to feed into Dietzenbach’s district network, serving over 2,000 households. Water work includes wetland protection with NABU in Hesse’s Büttelborn Bruchwiesen. Google reiterates its 24/7 carbon-free goal.

Office expansion includes Munich’s Arnulfpost for up to 2,000 staff, Frankfurt’s Global Tower space, and additional floors in Berlin. Local partnerships will fund digital skills and STEM programmes. Officials and customers welcomed the move for its benefits to infrastructure, sovereignty, and innovation.

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

Banks and insurers pivot to AI agents at scale, Capgemini finds

Agentic AI is expected to deliver up to $450 billion in value by 2028, as financial institutions shift frontline processes to AI agents, according to Capgemini’s estimates. Banks start with customer service before expanding into fraud detection, lending, and onboarding, while insurers report similar priorities.

To seize the opportunity, 33% of banks are building agents in-house, while 48% of institutions are creating human supervisor roles. Cloud’s role is expanding beyond infrastructure, with 61% of executives calling cloud-based orchestration critical to scaling.

Adoption is accelerating but uneven. Four in five firms are in ideation or pilots, yet only 10% run agents at scale. Executives expect gains in real-time decision-making, accuracy, and turnaround, especially across onboarding, KYC, loan processing, underwriting, and claims.

Leaders also see growth levers. Most expect agents to support entry into new geographies, enable dynamic pricing, and deliver multilingual services that respect local norms and rules. Budgets reflect this shift, with up to 40% of generative AI spend already earmarked for agents.

Barriers persist. Skills shortages and regulatory complexity top the list of concerns, alongside high implementation costs. A quarter of firms are exploring ‘service-as-a-software’ models, paying for outcomes such as the resolution of fraud cases or the handling of customer queries.

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