Tuesday, November 17, 2026 AI news, turned into opportunities PDF newsletter
AI Opportunity Daily
Subscribe
Health AIData centresSmart glassesRegulation

Apple studies how we ask AI about health, Telus quits two data centres, glasses get banned

Today's stories are about AI meeting the real world and the people in it. Apple has opened a year-long study of how people use AI for health questions, Telus has walked away from two planned Vancouver AI data centres after local opposition, and a growing list of courts, schools and governments are banning camera-equipped smart glasses.

Key takeaways

  • Apple is paying for evidence on how people use AI for health, and that evidence will shape what health AI is allowed to do.
  • Telus dropping two Vancouver AI data centres shows that local consent is now a real cost of building AI infrastructure.
  • Smart glasses are being banned place by place, which creates demand for rules, signage and compliant hardware rather than a single national law.
Story 1 of 3Tools4 min read

Apple opens a 12-month study of how up to 200,000 people use AI for health questions

Apple has released a free app that lets up to 200,000 US iPhone users ask health questions to an AI built with OpenEvidence while Apple studies how they use it. Apple says the data will not be used to train AI models.

Apple announced on November 15 that it is launching the Apple Health AI Usage Study, run in collaboration with the medical-search company OpenEvidence. The study is open to up to 200,000 participants aged 21 and over in the United States. To join, people need an Apple Intelligence-enabled iPhone, which Apple specifies as an iPhone 15 Pro or later, running iOS 26 or later. The app, called Apple Health AI Study App, is free in the App Store and is described there as a custom app built on technology from OpenEvidence.

Inside the app, participants can ask health questions and receive AI answers. TechCrunch reports that the answers are grounded in the New England Journal of Medicine, the Journal of the American Medical Association and Nature. Participants can also choose to share Apple Health data such as resting heart rate, activity, sleep and lab results, and they complete monthly surveys. TechCrunch says the study is meant to run for more than 12 months, so it is a long-term look at behaviour rather than a one-off demo.

The privacy terms matter because health data is sensitive. Apple says study data will not be used to train AI models. According to TechCrunch, Apple said the data is anonymized before it is sent to a secure third-party cloud provider, and that OpenEvidence deletes the query and response data after 24 hours. Apple vice president of health and fitness Sumbul Desai described the study as a way to understand how people engage with, understand and evaluate health AI in daily life. These are Apple's own descriptions of its safeguards, not independently audited claims.

The study arrives as general chatbots already field huge numbers of health questions, usually with no research behind how people act on the answers. Apple's approach is different in two ways: it picks a medical-literature tool rather than a general assistant, and it measures what users do and believe afterwards, not only what the model says. That second part is the scientifically valuable one, because the risk with health AI is often less about one wrong answer and more about how much people trust it.

There are limits. Participants are self-selected owners of recent, expensive iPhones in one country, so the findings will not represent everyone. Apple has not said when results will be published or whether the data will be shared with outside researchers. Still, the direction is clear: a large platform company is collecting evidence on health-AI behaviour that regulators, clinicians and app builders will likely cite. Anyone building in this space should expect that evidence to shape what counts as responsible design.

Why it mattersHealth is the most sensitive everyday use of AI, and the first large, well-funded evidence base on how people actually use it will influence design norms and rules. Builders who understand the findings early can design safer, more trusted products.

👀 What to watch

Whether Apple publishes interim results or opens the data to outside researchers, and whether competitors launch similar studies or tools.

3 opportunities from this story

1

Health-AI safety and disclaimer review for wellness apps

Medium⏱ 2–3 weeks💰 Fixed-fee audit per product, with an optional quarterly re-test retainer

Small wellness, fitness and telehealth apps are adding AI chat features faster than they are reviewing how those features answer medical questions. Offer a fixed-scope review that tests the chatbot against a list of risky prompts, checks the disclaimers and escalation messages, and produces a short report. Buyers are founders who worry about liability and app-store scrutiny, and Apple's study signals that how users trust health answers is now being measured.

Best for
QA testers, UX writers, clinicians turned consultants, compliance freelancers
First step this week
Draft a list of 30 risky health prompts and run them through one public wellness chatbot to build a sample report.
Open the full playbook
Launch steps
  1. Build a prompt set covering symptoms, medication, mental health and emergencies
  2. Score answers for accuracy, caution and clear escalation to a professional
  3. Write a one-page findings summary with suggested wording fixes
  4. Offer the audit to five small health or fitness app makers
Tools
Spreadsheet for scoringThe chatbot under testPublic clinical guidelines for referenceLoom for walkthroughs
Risks

You are not providing medical advice and must say so clearly; use a qualified clinician to review anything clinical. Avoid promising that an audit guarantees compliance.

2

Plain-language 'how to ask AI about your health' guides

Low⏱ 1–2 weeks💰 Per-workshop fee, per-guide licence, or a paid newsletter

People are asking AI health questions without knowing how to phrase them, what to verify, or when to see a doctor. Create short, accurate guides, videos or workshops for specific groups such as parents, older adults or employees of a company. Sell them to clinics, employers, community groups and insurers who want to educate their members.

Best for
Health writers, nurses, teachers, content creators
First step this week
Write a one-page guide titled 'Five rules for using AI for health questions' and show it to a local clinic or community group.
Open the full playbook
Launch steps
  1. Pick one audience and list their top 10 health questions
  2. Test how two AI tools answer them and note where they go wrong
  3. Turn the findings into a printable guide and a 30-minute talk
  4. Pitch it to a clinic, school or employer wellness programme
Tools
Canva or Google DocsAny major AI assistantZoom or in-person venueA qualified reviewer such as a nurse or doctor
Risks

Giving anything that looks like diagnosis creates liability. Keep the guide about how to use the tools and when to seek care, and have a clinician review it.

3

Privacy-explained summaries for health-AI apps

Medium⏱ 3–4 weeks💰 Paid newsletter or consulting briefs for organisations

Apple's launch highlights questions people rarely get answered: where does my health data go, who stores it, and for how long? Produce independent, readable summaries of how popular health-AI apps handle data, comparing their published policies. Sell a monthly briefing to employers, advocacy groups and journalists, or publish them as a niche newsletter.

Best for
Privacy analysts, lawyers, technical writers, researchers
First step this week
Compare the published privacy terms of three health-AI apps in one table and publish it as a free sample.
Open the full playbook
Launch steps
  1. Choose five apps and collect their privacy policies and any security documentation
  2. Define comparison criteria such as retention, training use and third-party sharing
  3. Publish the first comparison with clear dates and sources
  4. Offer a custom comparison to one organisation as a paid pilot
Tools
SpreadsheetNewsletter platform such as Substack or BeehiivArchive.org to date-stamp policies
Risks

Policies change and you must not claim an app is unsafe without evidence. State only what is documented and date every claim.

Story 2 of 3Business4 min read

Telus pulls out of two planned Vancouver AI data centres after local opposition

Canadian telecom Telus has dropped out of two AI data centres it planned with developer Westbank in Vancouver. Westbank says it will continue with a new partner.

Telus said on Sunday, November 15, that it is ending its participation in two planned AI data centres in Vancouver, according to The Globe and Mail. A Telus spokesperson said the company has decided not to proceed with co-developing the two previously announced data centres with developer Westbank. Telus gave no explanation in its statement. The Globe reports the decision followed heavy local opposition and recent shifts in the company's AI investment strategy, and cites an unnamed source who pointed to a lack of local support.

The projects were announced in May 2026 alongside the federal government as part of a sovereign AI infrastructure push. By the Globe's account, the plan called for about 63,000 Nvidia chips and up to 163 megawatts of electricity. One site was a converted office building in Vancouver's Mount Pleasant neighbourhood, expected to open at the end of 2026, and the other was a new downtown build due in 2029. The Globe adds that one of the sites is still subject to rezoning.

Opposition had been visible for months. Daily Hive reported earlier this year that a group called No AI Data Centres in Vancouver organised marches that drew hundreds of people, and a city councillor introduced a motion to pause consideration of large new data centres until the city has a framework for assessing community risks. Another councillor motion on stricter requirements is due in November, per the Globe. The new CEO, Victor Dodig, took over after Darren Entwistle retired in June, and the Globe frames the exit as part of a change of course.

Westbank has said it is proceeding with a new, unnamed partner, so the buildings may still happen. Telus says it will keep expanding capacity at its existing facilities, including one in Kamloops, B.C. That distinction is important: this is a retreat from two specific co-developments in a dense city, not a verdict on AI infrastructure in general. We do not know the real reason, and Telus has not given one, so any claim about its motives beyond what the Globe reports is speculation.

For anyone watching the AI build-out, the lesson is practical. Data-centre projects need power, chips, financing and also community consent, and the last of those is difficult to buy. Dense city sites raise questions about noise, energy use, grid capacity and who benefits. Expect more projects to face public hearings, and more developers to hire people who can explain a project to neighbours before the rezoning stage rather than after the protest.

Why it mattersIt shows that social licence is now a gating item for AI infrastructure, alongside chips and power. Projects in cities can slow or change hands even after a high-profile government announcement.

👀 What to watch

Who Westbank names as the replacement partner, how the November council motion is decided, and whether other telecoms or developers revise urban data-centre plans.

3 opportunities from this story

1

Community-consultation packs for data-centre developers

Medium⏱ 3–4 weeks💰 Per-project fee plus an hourly rate for meeting facilitation

Developers now have to win over neighbours as well as planners. Offer a service that prepares plain-language project explainers, FAQ sheets, noise and energy summaries, and a plan for public meetings. Buyers are developers, consultants and local authorities who need to show they listened before a rezoning hearing.

Best for
Communications freelancers, urban planners, community engagement consultants
First step this week
Write a sample two-page 'what a data centre means for your street' explainer for a fictional site and send it to three local developers.
Open the full playbook
Launch steps
  1. Collect common resident concerns from public council records about data centres
  2. Draft a template pack covering power, water, noise, jobs and traffic
  3. Practise a 20-minute public briefing and record a short sample
  4. Pitch the pack to developers with projects at the rezoning stage
Tools
Google Docs or CanvaPublic council meeting minutesA simple mapping tool such as Google My MapsZoom for virtual sessions
Risks

You must not mislead residents or hide trade-offs; clients who want spin will damage your reputation. Be accurate and let the numbers come from the developer's own engineering data.

2

Data-centre rezoning and council-motion tracker

Medium⏱ 2–4 weeks💰 Monthly subscription for professionals, with a free summary tier

Local rules on data centres are changing city by city, and developers, investors and journalists struggle to keep up. Build a simple tracker that logs rezoning applications, pause motions and new requirements in your region, with short summaries and dates. Sell access as a paid newsletter or a subscription dashboard.

Best for
Researchers, indie developers, journalists, policy analysts
First step this week
List every data-centre-related council agenda item from the past three months in your city in one sheet and share it free.
Open the full playbook
Launch steps
  1. Pick one region and subscribe to its council agendas
  2. Log each item with a status, date and a link to the source document
  3. Add a short plain-English note on what changed
  4. Send a weekly digest and charge for alerts and archives
Tools
Google Sheets or AirtableCouncil websites and RSS feedsNewsletter platformAn AI assistant for first-draft summaries, checked by you
Risks

Errors in legal or planning facts can mislead paying readers. Link every entry to the original document and correct mistakes publicly.

3

Energy and grid-impact explainers for AI projects

High⏱ 4–6 weeks💰 Commissioned reports and retained advisory work for councils or developers

Residents and councillors often cannot judge whether a figure like 163 megawatts is large or small. Create independent explainers that translate megawatts into household equivalents, show grid and cooling implications, and compare options. Sell them to councils, community groups and developers who want credible third-party material.

Best for
Energy analysts, engineers, data journalists, sustainability consultants
First step this week
Write a one-page explainer converting 163 MW into household and industrial equivalents, sourced to published grid data, and post it publicly.
Open the full playbook
Launch steps
  1. Gather local grid and utility data that is publicly available
  2. Build a simple calculator for load, cooling and carbon scenarios
  3. Have a qualified engineer review your method
  4. Publish a sample report and offer a paid local version
Tools
Spreadsheet or Python notebookUtility and regulator public reportsDatawrapper for charts
Risks

Technical mistakes will be noticed quickly. Use ranges, state assumptions, and get a qualified review before selling anything to a decision-maker.

Story 3 of 3Policy4 min read

Smart-glasses bans spread from New York courts to Australian schools and Punjab

Courts, schools, a military branch and a Pakistani province have each restricted camera-equipped smart glasses, mostly through separate local rules rather than one national law.

Engadget reported on November 15 that a growing list of places have banned AI-enabled smart glasses, meaning glasses with built-in cameras, microphones or computing. The best-known case is New York, where the state's Unified Court System has barred such glasses from all of its roughly 1,240 state and local courts since July 20, 2026. Engadget describes it as the first US state to ban the devices in every courthouse. The ban reportedly also covers prescription glasses that have recording features, and visitors must hand the devices in before entering.

Schools are the next front. Western Australia's education minister, Sabine Winton, announced a ban on smart glasses in public schools that took effect on October 12, covering wearing, using, charging and storing them on school property and at school events. The Next Web reports that the department had recorded no cases of irresponsible use, so the rule is preventive. Engadget adds that Victoria plans similar school restrictions from January 27, 2027. Exemptions exist in Western Australia for approved medical, accessibility and educational reasons.

Other restrictions come from different directions. Engadget reports that US Marine Corps policy at Camp Lejeune and other bases does not allow smart glasses. In Pakistan, the Punjab government banned the use and sale of camera-enabled smart glasses and similar gadgets for 30 days with immediate effect, according to Dawn on October 22. The ban covers retailers selling online or in stores, and law-enforcement users are exempt. The New York, Marine Corps and Victoria details rest mainly on Engadget's reporting, while the Punjab details come from Dawn.

The common thread is not a shared law but a shared worry: a camera that looks like ordinary glasses can record people without them noticing. Courts worry about witnesses and jurors, schools worry about children being filmed, and the military worries about sensitive locations. Because each owner of a space can set its own rule, the result is a patchwork. A pair of glasses that is acceptable at an office may be prohibited in the courthouse next door.

The patchwork has limits and unknowns. These are local, sector-specific rules, not national bans, and the Punjab measure is temporary. Enforcement is practical rather than legal: staff must spot the glasses, and the models vary in how obvious their cameras are. Manufacturers have not agreed on a standard indicator light or recording notice. Expect more institutions to copy the early adopters, and expect pressure on device makers to make recording visible.

Why it mattersWearable cameras are becoming ordinary, and institutions are writing rules before national lawmakers do. That patchwork creates compliance work for anyone who runs a public space, a school or an event.

👀 What to watch

Whether more US states or federal courts follow New York, whether Victoria's school rule takes effect on schedule, and whether glasses makers add standard recording indicators.

3 opportunities from this story

1

Smart-glasses policy templates for schools and venues

Low⏱ 1–2 weeks💰 One-off template bundle plus optional paid customisation

Schools, event venues, gyms and clinics are being asked about smart glasses but have no ready policy. Write short, editable policy templates, signage and staff scripts covering what is banned, what exemptions apply and how to handle prescription wearers. Sell them as a bundle to administrators who want to be ready before a parent or visitor complains.

Best for
Policy writers, education consultants, lawyers, freelance compliance advisers
First step this week
Draft a one-page template policy modelled on the published New York court and Western Australia school rules and send it to five schools or venues.
Open the full playbook
Launch steps
  1. Read the published rules and extract common elements
  2. Draft a policy, sign wording and an incident checklist
  3. Have a lawyer or education adviser review for your jurisdiction
  4. Sell on your site or through direct outreach to administrators
Tools
Google DocsCanva for signageGumroad or Stripe payment linkYour local education or court rules
Risks

Laws differ by place and you are not giving legal advice. Label templates as starting points and recommend local review.

2

Front-desk training for spotting and handling smart glasses

Low⏱ 2–3 weeks💰 Per-session training fee or per-venue licence for the materials

Staff at courts, clinics, schools and conference venues must now tell recording glasses from ordinary ones and handle the conversation politely. Run short training sessions and produce one-page guides showing common models, how to ask people to remove them, and where to store them. The buyers are security firms, venue managers and facilities teams.

Best for
Security trainers, hospitality consultants, corporate trainers
First step this week
Photograph or collect public images of five popular smart-glasses models and build a simple identification sheet.
Open the full playbook
Launch steps
  1. Create a spotting guide with model examples and common features
  2. Write a polite script for asking someone to remove or surrender glasses
  3. Run a free pilot session at one venue and collect feedback
  4. Package the session as a paid 45-minute training
Tools
Slides or CanvaPublic manufacturer product pagesZoom for remote delivery
Risks

Models change fast, so your guide goes out of date. Date every guide and refresh it quarterly, and avoid advising staff to use force.

3

Visible-recording accessories and prescription-lens services

High⏱ 6–8 weeks💰 Product sales to venues, or bulk contracts for storage and handling kits

Some wearers will keep using smart glasses where allowed, and venues want a way to show recording is off. Explore a small product or service line, such as clip-on indicator covers, lockable pouches for storage at entry, or cleaning and prescription-lens services for compliant frames. This is a hardware-adjacent idea best tested with a very small batch first.

Best for
Hardware hobbyists, small e-commerce sellers, optical shop owners
First step this week
Ask five court-adjacent venues or schools whether they would pay for lockable storage pouches, before you buy any stock.
Open the full playbook
Launch steps
  1. Survey venues about what they actually need at entry points
  2. Source samples of storage pouches or lockers from a supplier
  3. Test with one venue for two weeks
  4. Order a larger batch only after paid commitments
Tools
Alibaba or local suppliersShopify or a simple invoiceA small survey form
Risks

Inventory and shipping risk are real, and rules may change quickly. Pre-sell before buying stock, and avoid products that could be mistaken for tamper devices.

Get these as a PDF every 3 days

Free. One email every 3 days. Unsubscribe any time.

More editions