Subscribe

The AI intelligence layer

AI is moving fast.Stay ahead.

Daily AI news, clear explainers, and a living record of every model and milestone — for beginners and experts alike.

New to AI? Start here → Get the briefing →

Real developments.
Deeper perspective.
Brighter possibilities.

Model release Robotics Autonomous systems Research
AI PULSE
12:26A faster checkpoint for NVIDIA’s surgical video model12:26NVIDIA documents its open speech-to-speech model, weak spots included12:26Fifty models, 258 cognitive experiments, and a gap that is not closing fast12:26Agents can copy how an app looks. Finishing it is another matter12:26Qwen’s new image model is open-weight, not open-licensed14:16Google, Nvidia and Emerald AI launch an alliance on grid-flexible data centres14:16Google ships Gemini 3.8 Live for real-time voice agents14:16OpenAI publishes a youth safety blueprint for Australia12:26A faster checkpoint for NVIDIA’s surgical video model12:26NVIDIA documents its open speech-to-speech model, weak spots included12:26Fifty models, 258 cognitive experiments, and a gap that is not closing fast12:26Agents can copy how an app looks. Finishing it is another matter12:26Qwen’s new image model is open-weight, not open-licensed14:16Google, Nvidia and Emerald AI launch an alliance on grid-flexible data centres14:16Google ships Gemini 3.8 Live for real-time voice agents14:16OpenAI publishes a youth safety blueprint for Australia

What changed

Briefs, model releases, chip announcements and timeline entries, newest first.

More latest stories →

BriefA faster checkpoint for NVIDIA’s surgical video modelNVIDIA has added a distilled checkpoint to its open-weights surgical video model, claiming a near thirteenfold speed-up, with…BriefNVIDIA documents its open speech-to-speech model, weak spots includedThe technical report for NVIDIA’s open full-duplex speech model reports strong interruption handling alongside noticeably wea…BriefFifty models, 258 cognitive experiments, and a gap that is not closing fastA new framework re-runs 258 cognitive science experiments on 50 language models and finds their agreement with human answers…BriefAgents can copy how an app looks. Finishing it is another matterA new benchmark gives the leading model 58.1% overall but only 2.8% of tasks passing every test, which says something about h…BriefQwen’s new image model is open-weight, not open-licensedAlibaba’s Qwen team has published a 7B image generation and editing model with native transparency, but the weights carry a r…BriefGoogle, Nvidia and Emerald AI launch an alliance on grid-flexible data centresA new coalition wants data centres to adjust their power draw in response to grid conditions — the structure and intent are s…BriefGoogle ships Gemini 3.8 Live for real-time voice agentsGoogle’s new Gemini 3.8 Live audio models complete 35.1% of Sierra’s τ-Voice-banking tasks — the figure that says most about…BriefOpenAI publishes a youth safety blueprint for AustraliaA six-pillar policy document aimed at Australian regulators that restates protections OpenAI has already shipped and adds no…

The Canon

Evergreen explainers, maintained rather than published once.

All explainers

CanonWhat Is a Large Language Model? How LLMs Actually WorkA large language model learns to predict the next piece of text. That single mechanism explains both what these systems can do and how they charact…Reviewed September 2026CanonWhy AI Models HallucinateHallucination is not a bug in language models. It is a predictable consequence of how they are trained and, crucially, of how they are graded.Reviewed September 2026CanonAI Benchmarks Explained: What the Scores Actually MeasureBenchmark numbers drive coverage, procurement and valuations. Most of the widely-quoted ones are saturated, contaminated, or not comparable between…Reviewed September 2026

Who builds AI silicon

The companies designing the chips AI runs on — what they actually make, and whether you could ever buy it.

7 of 12 companies here design AI silicon you cannot buy.

See all 12 companies Or compare every chip

The AI Timeline

Key moments. A clearer picture.

Sep 17
2026 Policy
Policy

Anthropic lets vetted labs past its biology safeguards

Anthropic gives verified life science teams access to biology work its models otherwise refuse.

Sep 16
2026 Applications
Applications

OpenAI puts conversational ads inside ChatGPT

A US test lets users who click an ad talk to a business-sponsored agent first.

Sep 10
2026 Applications
Applications

OpenAI opens its agent harness as a managed API

The Codex harness behind ChatGPT and Codex enters public beta as the Agents API.

Sep 10
2026 Policy
Policy

Anthropic names the misuse operations it disrupted

A threat report details state-linked and criminal abuse of Claude across seven categories of harm.

Sep 5
2026 Policy
Policy

The DOJ backs OpenAI in the New York Times case

The US government argues transformative public benefit outweighs publisher harm.

Why it matters

We track AI developments not just to report what’s new, but to explain what it means — for industries, economies, and the future.

Our editorial approach →

The signal, not the noise.

A curated briefing on the week’s most important AI developments, delivered to your inbox.

Curatedby experts
Once a weekno more
No spamjust signal