In partnership with
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
November 2026 · Week 1
November 2, 2026
Good morning! This week, better AI is not just about producing more. It is about seeing where an answer came from, deciding what access to trust and knowing when new hardware is actually useful. Here are three developments worth your attention.
AI AT WORK
Google Sheets gives AI answers a paper trail
Original AI generated concept illustration, not a software screenshot.
Google announced on October 28 that Fill with Gemini in Sheets now attaches web sources to individual AI generated cells. Instead of checking a whole spreadsheet against a vague list of references, you can follow the source behind a particular answer. For anyone assembling research, comparing suppliers or sorting a working list, that is a more useful upgrade than another promise of effortless automation.
Google also expanded requests to up to 3,000 cells and reports faster processing. Its much larger speed claims come from an internal study, not a general guarantee for your work. The announcement says rollout began October 20 and can take up to 15 days. Availability depends on an eligible Workspace, consumer or education plan; a standard free Google account is not enough.
The important distinction is between traceability and correctness. A source link can make an answer easier to inspect without proving that the page is reliable, current or actually supports the cell. A cited mistake is still a mistake. For research that informs a purchase or a client recommendation, the judgment has not disappeared; it has become easier to exercise.
Our read: this is the direction everyday AI needs to go. We should be able to ask not only for a result, but also where it came from and what would change our mind. A spreadsheet full of polished answers is less valuable than a smaller set we can defend. Keep the source beside the answer, and the human decision beside both.
SECURITY AND MONEY
The Binance breach puts access under the spotlight
Original AI generated concept illustration, not an incident photograph.
Binance's October 29 breach has become a major test of operational trust. The exchange estimated an impact of around $6.3 billion. On October 31, the FBI attributed the theft to North Korea's TraderTraitor operation. The exchange's initial investigation suggests a compromised engineer's laptop allowed unauthorized withdrawals; the technical investigation is not yet a final public account.
As of November 1, Binance said it had replaced about 95 percent of the affected assets using company resources and was restoring withdrawals in phases. Replacing assets is not the same as recovering the stolen coins, and the company's assurances are not an independent guarantee. Elliptic's estimate of roughly $6.4 billion also illustrates why figures need attribution rather than false precision.
Nothing in the sources reviewed establishes that AI caused the breach. Automation of transactions does not prove an AI attack. The broader lesson applies well beyond crypto: a powerful system can still be undone by an endpoint with too much authority. For our work, the question is not just whether a platform looks secure, but what an individual device or account can actually authorize.
THE HARDWARE QUESTION
Apple's M6 iMac makes AI part of the upgrade pitch
Image: Apple Newsroom. Supplied for editorial coverage.
Apple introduced the M6 iMac on October 27, with US prices starting at $1,499 and availability from November 6. The 24 inch desktop keeps its familiar format while adding new processing power, WiFi 7 and Bluetooth 6. Apple positions its chip and Neural Engine as an upgrade for local AI models and creative work.
That story matters because AI is becoming a hardware purchase argument, not just another app subscription. But the announcement's performance multiples are Apple's own benchmark claims, with specific test conditions and comparison machines. They are not a promise that every export, research task or client job will get faster by the same amount.
Our take: a new chip is interesting when it removes a bottleneck you can name. If your current machine handles your work, the existence of a better one need not create an urgent buying decision. If you are hitting memory, rendering or local model limits, compare the exact configuration and software you actually use. Apple Intelligence availability also varies by language and region; some Siri AI features remain in beta. The useful upgrade is the one you can explain without the launch headline.
YOUR RESET
Before trusting an AI answer, granting an app access or replacing a computer, ask one question: what evidence would make this a good decision? Pick one answer, permission or bottleneck to check this week. Better technology should leave us with clearer judgment, not just more confidence.
Thanks for reading,
The Great Reset
This newsletter is educational and is not investment advice or a solicitation to buy or sell assets.
