The week in AI: October 26 to November 1, 2026
25 AI stories from October 26 to November 1, 2026, in plain English, with what each one means for you and 11 things to try. Starts with: Mistral releases Large 4 as a free open-weights AI model.
Compiled from Today in AI. Every story links to its source.
New things you can use
- Mistral releases Large 4 as a free open-weights AI model (October 29, 2026). Open-weights models let schools, governments and businesses run powerful AI on their own systems, so their data never leaves home. It also gives Europe, and countries like Canada, an alternative to US and Chinese labs.
- Google unveils Gemini 4 Carbon, now top of a major AI ranking (October 28, 2026). The top spot among AI models keeps changing hands every few weeks. The newest models are aimed at security first, which should help protect the software and websites everyone relies on.
- Google DeepMind unveils Genie 4, an AI that builds playable worlds (October 27, 2026). AI is moving from making pictures and videos to making worlds you can step into. That could change games, training simulations and how teachers show students things they cannot see in person.
Work and business
- Google, OpenAI, Microsoft and Anthropic back one AI training opt-out (October 30, 2026). Bloggers, schools, small businesses and publishers will soon have one simple switch to keep their work out of AI training, instead of chasing each company’s rules.
- Comcast cuts 900+ jobs while rolling out AI agents for customers (October 30, 2026). Big employers are reshaping customer service jobs around AI while hiring for different roles. Learning to work with these tools is becoming part of keeping a job, not just getting one.
- Experts followed an AI agent’s hidden lean 86% of the time (October 27, 2026). Having a person check AI work only helps if their judgment stays their own. When you use AI for a decision, notice where it nudged you.
- Microsoft offers buyouts to about 60,000 staff, including in Canada (October 26, 2026). Big tech firms are reshaping their workforces around AI, and buyouts are becoming a quieter alternative to mass layoffs. If your own employer starts restructuring, it pays to know your options early.
Learning and school
- Microsoft school AI privacy rules backed by teachers’ unions begin (November 1, 2026). This gives teachers and parents a clear checklist for what school AI should and should not do. Canadian schools and boards can use the same questions when choosing AI tools.
- US computer science enrollment falls 12% as students change course (October 29, 2026). Students are already rethinking which careers AI will change. Coding still matters, but pairing it with another field, and with knowing how to work alongside AI, is becoming the safer bet.
Heads up
- ChatGPT now limits paid accounts per device and payment card (November 1, 2026). Families or small teams sharing one card or computer for several ChatGPT subscriptions could get locked out. A Business plan may be the better fit for teams.
- Study finds nearly 40% of AI chatbot election answers had errors (November 1, 2026). AI chatbots are handy, but on dates, deadlines and rules they can be confidently wrong. For anything official, including Canadian elections, check the election agency’s own website.
- A shared ChatGPT link could have taken over accounts, now fixed (October 31, 2026). Links to AI chats can carry risks like any other link. Be careful opening shared chats from people you do not know.
- ChatGPT to limit weapons info for teens and end violent chats (October 30, 2026). AI chatbots are adding more guardrails for young users. Parents and teachers should know what ChatGPT will and will not do for teens, and that parental alerts now exist.
- Fake Claude Code update in Cursor’s marketplace planted a backdoor (October 30, 2026). Even official-looking add-ons can be swapped for malicious ones. If you use AI coding tools, keep an eye on updates and act fast when a warning goes out.
- China-linked hackers used DeepSeek-powered AI agents on 200+ targets (October 29, 2026). AI agents let one group of attackers hit hundreds of targets at once. For any business, basics like strong passwords, multi-factor sign-in and prompt updates now matter even more.
- Anthropic AI built hacks for Chrome, Safari, Windows, Android, Linux (October 28, 2026). AI can now find serious security holes in everyday software quickly, for defenders and attackers alike. Keeping your browser, phone and computer updated matters more than ever.
- Google brings ads in AI Mode search answers to more countries (October 28, 2026). AI search answers will increasingly carry ads, so it is worth noticing what is sponsored. For small businesses, AI Mode is becoming a new place to be found, paid or not.
- Fake ChatGPT and Claude desktop apps on GitHub hide malware (October 27, 2026). People searching for AI apps are now a target. Download ChatGPT, Claude or any AI tool only from the company’s own website or official app store.
- Planted instruction files can trick AI coding agents into malware (October 26, 2026). An AI agent will follow instructions it finds in a file, even ones a stranger left there. If you use coding agents, where you start them matters.
Rules and government
- NAACP pollution lawsuit over xAI’s Memphis data centre can proceed (October 31, 2026). AI data centres need huge amounts of power, and neighbours are going to court over the pollution that can come with it. The outcome could shape how quickly new data centres get built.
- UK budget promises AI data centres faster grid hookups, with strings (October 29, 2026). Countries are competing to host the data centres AI needs, and communities want something back. Expect similar deals to be debated in Canada as projects grow.
- China orders job-impact reviews before some AI projects go ahead (October 26, 2026). Governments are starting to treat AI’s effect on jobs as something to check before deployment, not after. Expect the same question to come up in Canada and elsewhere.
Phones and devices
- Alibaba says Qwen will power Apple Intelligence on iPhones in China (October 31, 2026). AI features on the same phone can come from different companies depending on where you live. It shows how local rules decide which AI people get.
Science and discovery
- Anthropic buys one-person drug startup, shares Claude pharma results (October 27, 2026). A single person with AI ran a drug programme that would normally need a team. Paperwork-heavy jobs in science, health and manufacturing are where AI is already saving the most time.
- OpenAI says its AI disproved a major computer science hypothesis (October 26, 2026). If the claims hold up, AI is starting to produce original results in fields where progress usually takes years. For now they are claims, and the expert checking is what will tell.
Try this week
Using AI for a deadline or rule? Make it show its sources.
Answer this, then list the official government or election-agency web pages I should check to confirm it, and tell me which parts of your answer you are least sure of: [your question].
Teacher or school leader? Turn these rules into questions for any AI vendor.
I am evaluating an AI tool for my school. Write 10 plain-language questions to ask the vendor about student data, AI training, emotional attachment features, monitoring, human oversight and deleting data, and say what a good answer to each looks like.
Parent or teacher? Get a plain guide to teen safety settings.
Explain, in plain language, the teen safety features and parental controls available in ChatGPT and other popular AI chatbots, how to turn them on, and 5 questions I should talk through with a [age]-year-old about using AI safely.
Run a website? Ask AI to explain your options before you change anything.
Here is my website’s robots.txt file: [paste]. Explain in plain language what it currently allows, and show me how I would add the new __usage_ai-training opt-out without blocking Google or Bing search.
Run a small business? Get a quick security check-up plan.
I run a [type of business] with [number] staff using [email, accounting and other tools]. Give me a one-page checklist of the 10 most important security steps, like passwords, multi-factor sign-in, updates and backups, in plain language, most important first.
Choosing a program or career? Ask AI to stress-test your plan.
I am a [grade or year] student interested in [field]. List the jobs this leads to, how AI is likely to change each over the next 5 to 10 years, and which skills or second subject would make me more resilient. Be honest about uncertainty.
Not sure your devices are up to date? Get a quick checklist.
I use [list your devices and browsers, e.g. Windows laptop, iPhone, Chrome]. Give me step-by-step instructions to check each one is fully updated and to turn on automatic updates.
Using AI for a decision? Ask it to argue the other side too.
Here is the decision I am weighing and your earlier recommendation: [paste]. Now make the strongest case for the opposite choice, list the assumptions behind each option, and tell me which facts would change the answer.
Got a long technical document to check? Ask AI to review it alongside you.
Review this document for errors, inconsistencies, missing steps and unclear wording: [paste]. List each issue with the section it is in, why it matters and a suggested fix, most serious first.
Weighing a buyout or severance offer? Get your questions straight first.
My employer is offering a voluntary buyout. Help me list the questions to ask HR and a lawyer about pay, benefits, pension, taxes and timing, and build a simple table comparing staying versus leaving over the next two years. I am in [province or country].
Using a coding agent? Never start it in your Downloads folder, and read any instruction file you did not write first.
Here is an AGENTS.md or CLAUDE.md file I did not write: [paste]. Flag any instruction that installs packages, adds servers, changes settings or API addresses, or turns off approvals, and explain the risk of each in plain language.
More weeks
- Next: The week in AI: November 2 to 8, 2026
- Previous: The week in AI: October 19 to 25, 2026
- All editions of The Signal
Related Power of AI pages
Keep reading with Start here, everyday uses, the tool guide, the writing workshop, and sources and standards.