From the archive
Follow the latest AI news and artificial intelligence developments across models, agents, robotics, research and the wider industry. Explore new AI tools, advances in computing, business moves, safety and policy, with clear reporting and useful sources. Read the stories, catch up with News in Brief and join the conversation.
Lead story
The Dutch government is developing DAWO, a NixOS-based digital work environment intended to cover everything from operating systems and office software to cloud services and AI. Eight municipalities are already running trials, and the project is a bid to reduce reliance on technology suppliers that governments may not…
Read the report & discussion ↗MIT CSAIL Watch
MIT researchers used a machine-learning algorithm to find a formulation that kept RNA vaccine particles stable for up to a year at room temperature, or two months at nearly 37°C. In tests in mice, the stored vaccines produced immune responses comparable to a similar Moderna-style formulation, a result…
NVIDIA Watch
MeetKai says it will build sovereign AI systems in six countries using NVIDIA infrastructure, with a national-scale data centre planned for Ukraine from 2027. The plan puts locally hosted models and computing capacity on the table, although it remains a company announcement rather than evidence that the facilities…
DeepSeek Watch
DeepSeek has reportedly passed a $1bn annualised revenue run rate after raising API prices, while putting more than 70% of its computing capacity into training new models. The account suggests a shift from bargain-priced challenger to commercial operator, with a sizeable bet still being placed on what comes…
Hugging Face Watch
A team writing on Hugging Face says it has made LTX-2.5, a 22-billion-parameter audio-and-video model, generate five-second clips faster than they play on a GPU with 32 GB of memory. The practical win is not just speed: the authors describe techniques for fitting real-time talking-character generation onto more…
Hugging Face Watch
Adalat AI has released SCRIBE, an open-source tool that breaks speech-recognition errors into categories such as words, numbers and punctuation. Its central point is practical: one overall accuracy score can make a harmless spelling difference look as bad as a wrong medicine dose.
Watch Desk
An essay published on 28 September argues that AI products meant for serious work should help people check their answers, show detailed citations and use interfaces suited to the task, rather than defaulting to chat. Its sharper point: a convincing conversational style is not the same thing as…
Watch Desk
NaiveAI has released Naive-N0.5-Flash, an open-weight model with a native one-million-token context window and an MIT licence. The release is notable for putting a very large context claim alongside an accessible licence, though the supplied report gives no independent performance results.
Watch Desk
A METR research team built and deployed a basic monitor to flag harmful actions by AI agents during evaluations, then found serious problems when it assessed how well the monitor worked. The findings matter because monitoring is often treated as a safety backstop; METR’s account points to gaps…
Watch Desk
Unsloth’s latest release lets users run Qwen-Image-2.1 locally, create reusable agent skills and edit images as well as generate them. It also improves reasoning playback and adds practical changes to projects, model storage and Linux installation.
Watch Desk
AI products are racing from answering prompts to taking responsibility for long-running tasks, often with their own cloud computers and the ability to act across apps. A 36Kr analysis describes the shift, and the messy product-design question underneath it: how much autonomy is useful before people lose track…
Google DeepMind Watch
Google DeepMind engineer Kevin Hou argues that AI agents could make fixed, human-designed interfaces a thing of the past. His examples range from teams of AI agents building an operating-system kernel to researchers automating much of their analysis workflow, offering a striking look at how model capability could…
Yoshua Bengio Watch
AI researcher Yoshua Bengio argues that developers may not fully understand what advanced language models absorb from training data or what their training rewards. His warning is not that models have human intentions, but that systems can learn behaviour their makers did not mean to encourage, with growing…
Watch Desk
AI is advancing faster than governments can settle how to regulate it, and enforcement of the EU AI Act is already lagging. The New York Times reports that European Commission president Ursula von der Leyen convened senior officials near Brussels this month as regulators weigh AI’s promise against…
Meta AI Watch
Meta’s Muse AI agent can help users identify and cancel recurring subscriptions, putting a common household chore in the path of an assistant designed to handle everyday tasks. That could make it harder for subscription businesses to rely on people forgetting to cancel, though the immediate evidence is…
Watch Desk
A new analysis compares the estimated compute cost of AI systems with the time people spend on a database of tasks, arguing that AI can be much cheaper when it can perform the work. The striking implication is not that AI can do every job, but that low…
NVIDIA Watch
Researchers can now explore AI-predicted protein-complex structures for more than 2,800 viruses, alongside an open NVIDIA pipeline for making further predictions. It is a useful head start for outbreak research, not a shortcut around the laboratory bench.
Watch Desk
Cloudflare’s 2026 founders’ letter argues that growing AI-agent traffic could overwhelm websites and squeeze smaller publishers and businesses. The company says it is introducing tools to help crawlers fetch only updated content, alongside ways for creators to earn money when agents use their work.
Anthropic Watch
Anthropic has launched Claude Marketplace, bringing more than 2,000 connectors and plugins together with Claude-powered partner products and deployment services. It gives organisations a central place to find tools for Claude, rather than assembling integrations one by one and hoping the paperwork behaves.
Microsoft AI Watch
Microsoft says it helped dismantle an AI-powered phishing service that compromised more than 12,000 mailboxes across over 10,000 organisations. The company says criminals used the system to analyse stolen email and craft follow-up fraud, while two suspected operators were arrested in the UK.
