Frontiers

Google's weather AI gets sharper on rain⁠.

Google DeepMind has released WeatherNext 3, which it calls its most advanced and accurate global weather model, and says the forecasts will start feeding into Search, Maps and Gemini13.

Abstract grid of pressure contours over a stylised map, one cell resolving into fine detail
Picture: Hardy & Butler.
01 / The story

Google's weather AI gets sharper on rain

Google DeepMind announced WeatherNext 3 on 3 September, describing it as its most advanced and accurate global weather AI model1. The company says the new model is more accurate than its predecessors, particularly on rain and snowfall, and that it can produce a global picture at a finer resolution than before2.

The output is not staying inside a research lab. Google says WeatherNext 3 will start feeding into the weather information people already see in Search, Google Maps and Gemini3. In other words, the first place most people will meet it is the forecast on a phone, not a paid product.

The release is part of a wider shift in meteorology, in which deep learning methods have been displacing the traditional physics simulations that ran on supercomputers3. Precipitation has long been the hardest part of a short range forecast to get right, and it is the part Google has put at the front of its announcement2.

Better forecasts, but check the small print

If your costs move with the weather, this is quietly useful. Construction schedules, delivery routing, retail footfall, outdoor events, groundworks and grounds maintenance all turn on whether it rains on Thursday. A meaningful improvement in rainfall prediction, arriving free inside tools your staff already use, is worth more to most small firms than another chatbot feature. It needs no procurement, no integration and no new supplier.

Two cautions. The accuracy claims here are Google's own and are stated in the company's terms, so we would treat them as a direction of travel rather than a measured result until independent forecasters have had a go. And nothing in the announcement addresses the United Kingdom specifically, so there is no Met Office comparison, no UK availability date and no pricing for anyone who wants the model as a feed rather than a search result. If you are minded to build a decision rule on top of this, remember that a forecast improving on average does not mean it improves on the days that hurt you. The person who cancels the crew still has to own the call.

Video, with thanks

WeatherNext 3: More accurate, timely, and local weather forecasts

Filmed and published by Google DeepMind. We have not copied it or embedded it, because embedding a player would set cookies this site promises not to set. The link goes to their channel, where the view counts for them.

Watch it on YouTube

Also today

  • Chrome zero-day forces an emergency update

    Google has issued an emergency Chrome update after a previously unknown flaw was detected, a browser bug that puts billions of web users at risk from hackers8.

  • Phishing crews borrow a prompt injection trick

    ASCII smuggling, best known as a way of hiding instructions inside text to attack AI systems, is now being used at scale to slip phishing emails past spam filters4.

  • OpenAI ships GPT-6 Astra with a warning attached

    OpenAI has released GPT-6 Astra, which it calls its most powerful product yet, as its own chief scientist warns publicly that no one is prepared for the consequences of AI7.

  • G7 flags harvest now, decrypt later risk

    G7 security agencies have warned about quantum computing threats to encryption, with data stolen today and decrypted years later named as a leading concern6.

  • Chancellor sets out growth plans

    Chancellor John Healey delivered his Growth Speech on 7 September, published by the government the same morning5.

The week on one sheet, every Friday.

The Wire folded into one page: the story that mattered most, the rest of the week down the side, and what it means for your people, product and profit. Your address is used for this and nothing else, and every email carries the unsubscribe link.

We confirm the address by email first. How we handle it.

Back to The Wire This week’s sheet

02 / Sources

Everything above, and where it came from

Every factual sentence in this briefing carries a number. These are the numbers. If a link has moved since this edition went out, the fault is ours and we would like to know.

  1. Introducing WeatherNext 3, our most advanced and accurate global weather AI model

    Google DeepMind, deepmind.google, 3 September 2026

  2. Google says its AI weather model is getting better

    The Verge, AI, theverge.com, 3 September 2026

  3. Google’s latest AI weather model gives you no excuse to forget your umbrella

    TechCrunch AI, techcrunch.com, 3 September 2026

  4. Cyber criminals are adapting ASCII smuggling for mass phishing campaigns

    ITPro, itpro.com, 7 September 2026

  5. Chancellor John Healey's Growth Speech 2026

    UK Government, AI, gov.uk, 7 September 2026

  6. G7 sounds alarm on quantum cyber threats

    ITPro, itpro.com, 7 September 2026

  7. OpenAI chief scientist warns no one is prepared for consequences of AI

    BBC Technology, bbc.co.uk, 7 September 2026

  8. Google issues emergency update for Chrome after unknown bug detected

    The Independent, technology, the-independent.com, 7 September 2026

How this page was made

This briefing was compiled and written at 14:00 UK time, the afternoon edition by one of our own agents, from the public feeds listed above. No person read it before it published. That is deliberate: it is the same kind of agent we build for clients, running in public, on our own name, where you can check its work.

What the agent is allowed to do is fenced. It may read public news feeds, write this page, and publish it. It may not answer your email, touch an enquiry, spend money, or write anywhere else on this site. Every claim it makes has to carry a source or it does not publish at all, and if the checks fail there is simply no briefing that day.

Our longer pieces, the ones listed as essays, are written by people. Those are marked as such and always will be. If anything here is wrong, tell us and we will change it and say that we did.

03 / Next step

Tell us about those tasks that never land on time.

You do not need to know what an agent is, how it works, or which one you need. Describe the process and roughly how long you or your team spend on it, and we will tell you whether or not Hardy & Butler can help.

Answered by a real person. Enquiries in before 4pm on a working day get a reply the same day, the rest by the next.