AIIntelGet the Briefing

Latest

NewsletterNo. 1

AIIntel Weekly Briefing No. 1

12 – 19 October 2026

Issued

This week

Governments and parliaments account for much of this week. The Canberra Times reports that the Australian Government does not support the Community and Public Sector Union's claim for a standalone condition on AI in public-service bargaining, including protections against using AI for surveillance, performance assessment, recruitment or promotion. Researchers told Australia's parliamentary AI inquiry that the Australian AI Safety Institute is not resourced to be taken seriously internationally. Kenya's National Assembly has given its communications committee one month to report on how far AI has spread into academic and health sectors, as lawmakers consider a standalone law. Indonesia's Ministry of Communication and Digital Affairs is studying a regulation that would require developers to label AI-generated content. Morocco's new government, appointed on 16 October, includes a Minister of Artificial Intelligence, Digital Transformation and Administration Reform. The Guardian reports that Austrian MEP Lena Schilling has urged the EU to make a December ban on nudifier tools enforceable, after sexualised deepfakes of her circulated online.

Company announcements sit beside those steps. Anthropic said that, subject to stockholder approval, it is committing 5% of its shares to a new Anthropic Public Benefit Fund, and that a further 2% may be added, at the board's discretion, for every 50% increase in the company's value above a set baseline. Microsoft has ended its agreement for the AI data centre Nscale is planning at Loughton in Essex, a project the UK government had described as the largest UK sovereign AI datacentre. OpenCode said a security breach it linked to a distillation attack affected 1.4% of users of its Console service, a subset who had external providers linked. EverWind announced plans for a data-centre campus at Point Tupper in Nova Scotia, and CBC News reports that Anthropic will lease the first phase.

Regulation

AI news

Plus 37 more stories

Tracker changes

26 more in the tracker

Products and releases

  • DeepSeek

    DeepSeek-V4.1-Flash-1017

    Upgrade to V4.1 Flash on the DeepSeek API, web and app; API model name (deepseek-flash) and pricing unchanged. DeepSeek also opened its in-house search API.

    Source (opens in a new tab)

  • NVIDIA

    RTX Spark laptops

    Windows laptops built on a single superchip combining an NVIDIA Blackwell RTX GPU and an NVIDIA Grace CPU, with up to 128GB of unified memory and up to 1…

    Source (opens in a new tab)

  • Elorian

    Glimpse (research preview)

    Research preview of a visual thinking model that zooms into images, rewatches video at denser frame rates and uses tools that write code while it reasons;…

    Source (opens in a new tab)

  • Reflection

    Beam 1.0

    First open-weight model: 501B-parameter mixture-of-experts (23B active) with Apache 2.0 weights on Hugging Face, plus Reflection's API in public beta.

    Source (opens in a new tab)

  • Kuaishou (Kling AI)

    Kling 4.0

    New-generation video model launched on 15 October: native single-pass clips of up to 30 seconds, steadier large-scale motion and camera work, and Omni…

    Source (opens in a new tab)

  • StepFun

    Step 5 Preview

    Open-weight flagship model for agentic work: a sparse mixture-of-experts model with 600B total and 27B active parameters, up to 1M tokens of context and…

    Source (opens in a new tab)

  • Institute of Foundation Models (MBZUAI) and Inception42

    Jais 3

    Six open language models, from 900 million to 375 billion parameters, plus Jais Speech for Arabic and dialect speech recognition. Built on IFM's K2 Horizon,…

    Source (opens in a new tab)

  • Mistral

    Timestral

    First time-series foundation model family, in four sizes (5M, 28M, 123M and 209M parameters). Mistral says the 209M model ranks first on GIFT-Eval, TIME and…

    Source (opens in a new tab)

8 more products

Coming up