Kia ora! Welcome to New Zealand’s weekly roundup of AI news, events, jobs and education.
It’s Monday.
Let’s start the week with something light (and this video is well worth your time): a glimpse into a future where you use ChatGPT for everything 😂
Links:
This week’s highlights
⚠️ Meta scraping exposes NZ’s AI protection gap
🚓 Whanganui firm’s AI catches burglars
💬 John Kirwan-backed AI therapy chatbot
💼 Autohive pitches AI ideas to mayoral candidate
🎓 Auckland Uni’s AI assistant handles 9,000 chats
🎙️ Gavin Sharkey: fixing AI pilot purgatory
🏡 AI at Home: ChatGPT projects for home admin
📚 AI Leaders Relevance has an expiry date
👷 AI tool to check dodgy builders
🦜 Victoria Uni trains AI to identify kākā
💡 Tip: Turn NotebookLM into your strategy co-pilot
Happy reading and listening ✌️
Did someone forward you this? Sign up!
HEADLINE STORY
⚠️ Meta scraping Kiwi data exposes NZ's AI protection gap
Source: Getty.
📣 Word On The Street: AI experts warn we're near last globally on the protection spectrum as Meta scrapes Kiwi data without consent since 2007.
🔎 Zooming In:
Meta trained AI on NZ Facebook and Instagram photos since 2007, including children's images.
80% of Kiwis want AI regulated, yet government shows minimal interest in protection frameworks.
Experts propose risk-based approach: ban surveillance tech, regulate healthcare AI, and allow low-risk uses.
🏘️ Our Take: Every powerful technology gets guardrails. Speed limits for cars. Safety catches for guns. Building codes for construction.
AI is the glaring exception in NZ. Meta just admitted to scraping 18 years of Kiwi social media data for AI model training.
The EU bans high-risk AI like social scoring. The UK enforces five core principles through existing regulators. Australia prepares sector-specific reforms. Even the US, despite Trump's deregulation push, sees states like California mandate safety disclosures. New Zealand layers codes onto existing privacy laws whilst ranking near bottom globally for AI trust. Surely no coincidence…
Dr Andrew Lensen's proposed framework provides a form of regulatory logic.
Ban the dangerous stuff like facial surveillance and state tracking outright.
Regulate the middle ground where harm could occur: social media AI and healthcare applications need guardrails.
Let low-risk tools like email summarisers run free without red tape.
The conversation at the NZ Government level needs to accelerate as AI tools proliferate and risks to data privacy and handling in NZ enter new territories.
Master AI. With people who get it.
A practical Masters for working professionals. Learn in trusted cohorts, apply it at work.
academyEX is the EXceptionally proud sponsor of this week's newsletter.
💼 Business & Industry
Wellington mayoral candidate turned to Autohive (AI Agent Platform) for AI ideas. Dan Milward brought the local AI firm into WCC brainstorming workshops to solve legacy tech problems. Autohive's suggestions: automated LIM processing, predictive parking analytics, and AI call systems that replace outdated council software without massive capital spend. The pitch centred on speed: prototype solutions fast instead of multi-year procurement cycles.
3-min read.
Our take: This is the first time I’ve seen local political candidates asking tech people (let alone startups) for ideas. It’s a refreshing shift. These people sit at the intersection of actually understanding Wellington’s landscape, economy, and population, and the technology solutions to improve outcomes. If the Council implemented even 10% of the use cases recommended by Autohive, Wellington would be a better place. This kind of blueprint is exactly what more politicians should be looking at: progressive thinking from organisations that move fast, use human resources efficiently, and know how to turn innovation into impact.
Auckland Uni's AI chatbot handles 9,000 monthly conversations. Three years of pilots with IBM Watson and local startup ElementX AI, with zero promotion, yet staff and students now prefer the AI Assistant over phoning campus. It's fielding 60,000 searches monthly across 90+ digital domains in multiple languages.
3-min read.
Our take: The real story isn't automation or AI, it's proving organic adoption beats forced rollouts. When AI genuinely solves problems, users find it themselves, which is the lens and benchmark other institutions and businesses should be thinking from.
Whanganui security firm's AI catches 508 burglars and wins national award. Alarm Watch upgraded its 2008 app with AI that identifies people, guns, fires and loiterers from live camera feeds. The system filters out dogs and false alarms before staff even see them. 35,000 monitored customers now get verified alerts instead of notification spam.
2-min read.
Our take: The business model evolution here is a playbook for local firms. Alarm Watch has turned its monitoring staff from alarm checkers into threat responders. That’s the difference between reacting to 100 alerts and acting on 10 verified incidents. The technology is making them far more effective at the only job that matters: stopping actual crime. This is a glimpse into what happens when workers focus on the application of AI rather than concerning themselves with the job loss narrative associated with AI.
58% of Kiwi workers have received zero AI training. Dayforce surveyed 520 New Zealanders. Only 19% say AI is a career priority, yet 52% of companies offer no training courses at all. Executives are 30% more likely than frontline staff to believe their company uses AI responsibly, and three-quarters say of execs they're ready for AI. Less than one in four frontline workers agree.
2-min read.
Our take: The readiness gap isn't about perception, it's about who gets resources. Executives feel prepared because they're getting training and context but frontline staff don't trust AI because nobody's shown them how it works. Companies are creating a two-tier workforce: those driving AI innovation and those scrambling to keep up.
💼 Government & Legal
Kiwi entrepreneur built AI tool after $1M dodgy builder disaster. CheckMyBuilder scans court filings, company registers and liquidation notices for $49, flagging builder insolvency patterns humans miss. Over 1,000 reports generated with zero advertising shows massive demand for accessible due diligence.
3-min read.
Our take: Building a home is most families' biggest investment: one bad decision means years of stress and legal battles. AI is democratising due diligence that was previously only accessible to those who could afford lawyers. This solves a real market gap: fragmented public data that's technically accessible but practically invisible. These problems to be solved are hidden in plain sight.
💼 Education & Society
Health NZ just launched an AI therapy chatbot backed by Sir John Kirwan. Ask Groov provides 24/7 mental health support for mild-moderate cases, and anonymises user data and pulls from psychiatrist-written content rather than the likes of Reddit. But child psychiatrists question English-language bias affecting Māori patients and lack of longitudinal clinical testing before public launch.
1-min read.
Our take: Teens are using AI for therapy whether we like it or not. Creating safer alternatives makes sense, but without proper testing on vulnerable populations, we're experimenting in real-time with real lives. This is a step-up on ChatGPT therapy and should be monitored closely
Victoria University trained AI to identify individual kākā from photos using beak patterns and scratches. Researchers set up GoPro feeders in Zealandia (Wellington) where birds stick their heads in for nectar and get photographed. The pilot worked in controlled conditions with white backgrounds. Next challenge: making it work across Wellington with varying backgrounds and angles instead of photo-booth style setups.
2-min read.
Our take: This flips conservation's fundamental constraint. Bird populations were limited by how many researchers could physically band and track. Now the limit is camera placement, not human capacity. That's how AI application should be thought about; not incremental improvement but removing bottlenecks completely.
🎙️ The AI Corner Podcast
This week’s guest is Gavin Sharkey, Managing Director of Kōwhai AI. Hear:
Why New Zealand businesses keep stalling after AI pilots, and how to fix it.
How to build internal capability instead of relying on consultants.
Gavin’s take on AI platforms, innovation culture, and the real risk of overbuilding.
💡 Tip of the Week: Turn NotebookLM Into Your Strategy Co-Pilot.
Brought to you by academyEX.
NotebookLM isn’t just for summarising notes. It’s a thinking partner that helps leaders see around corners:
Uncover hidden risks in reports
Surface growth opportunities in team updates
Draft board-ready or exec-ready summaries
Challenge assumptions in planning docs
Turn raw insights into concrete actions
How to use it:
Create a new Notebook.
Upload your sources (reports, meeting notes, strategy decks, URLs, audio, etc.).
In the chat box, prime it with: “Act as my strategic advisor…”.
Then ask: “What risks, opportunities, or blind spots stand out?”.
Export insights into a simple Now / Next / Later plan of opportunities, and pressure-test decisions before they hit the real world.
Pro tip: Keep separate notebooks for key themes, such as Innovation, Market Trends, or Risk Signals. Add more sources over time and you’ll build a searchable archive of your own strategic intelligence.
Example of a Notebook set up for AI trends.
📚️ AI for Business
Helping leaders and teams adapt, learn, and scale with AI.
1️⃣ AI Leadership: Relevance Has an Expiry Date: For too many execs, “AI adoption” still means using Copilot at work. Younger teams are stitching 10-30 AI tools a day into workflows that move faster than their incumbent competition.
4-min read.
2️⃣ Is ChatGPT (demo link below) Becoming the West’s First Super App?: Spotify, Canva, Booking.com, Figma, Coursera, and more now live inside ChatGPT. Could this be the start of the AI era’s WeChat moment where every app runs through a single conversation?
4-min demo.
3️⃣ Agents Are Here To Work, Not Chat: The new generation of AI tools isn’t about conversation, it’s about delegation. Agents act on your behalf, finishing the task while you focus elsewhere.
1-min read.
4️⃣ 5 Pieces of AI Content to Review This Week:
1) It’s Google’s AI lead to lose,
2) OpenAI’s next monopoly move,
3) the two drivers fuelling global AI growth,
4) the AI gold rush pattern we should monitor, and
5) updated Will Smith spaghetti benchmark.
3-min scroll with links to each story.
🏡 AI at Home: Organise Real-Life Projects with ChatGPT
ChatGPT Projects are smart, organised spaces that keep everything for your home life in one place. They’re ideal for managing renovation plans, meal prep, budgeting, travel, and family routines. Each Project keeps its own memory, instructions, and files, so ChatGPT remembers the context, knows how to help, and builds on your work over time.
Each Project’s instructions define ChatGPT’s role, tone, and output style, and they persist across every chat in the project, so it always knows how to respond without reminders from you.
The memory in each Project keeps context, files, and past messages separate, with the option to stay isolated or draw from your broader account memory.
You can attach files to any Project to turn it into a real workspace, letting ChatGPT reference and build on your documents every time you return.
Example Project Folder ideas:
Home admin: Track bills, repairs, and maintenance.
Family routines: Plan meals, chores, or weekly schedules.
Finances: Budgeting, investments, or retirement planning.
Travel: Build itineraries, budgets, and packing lists.
Events: Coordinate birthdays, holidays, or renovations.
🌍 Tech Updates From Global
The selected top headlines from each major AI tech company.
OpenAI
Acqui-hired personal finance assistant Roi to embed financial reasoning into ChatGPT and develop adaptive consumer-facing tools.
Unveiled Agent Builder and AgentKit at DevDay for drag-and-drop AI automation and agent deployment.
Launched Apps SDK allowing third-party apps like Spotify and Figma to run inside ChatGPT.
Partnered with AMD to secure 6GW compute capacity 10% company stake.
Expanded ChatGPT Go to 16 Asian countries with a sub-$5 plan.
Launched Gemini 2.5 Computer Use model outperforming OpenAI's agent on web/mobile benchmarks.
Introduced Gemini Enterprise platform ($30/user/month) and CLI Extensions for building workplace AI agents and command-line tools.
Released PASTA for user-adaptive creative tools and Prompted Variants for dynamic UI.
Anthropic
Launched Claude Code plugin support in public beta for custom agent development.
Partnered with Deloitte to deploy Claude across 470,000 employees and IBM to integrate Claude into enterprise software.
Welcomed former UK Prime Minister Rishi Sunak as senior adviser.
Microsoft
Added Rishi Sunak as adviser on global AI strategy.
Amazon
Launched Quick Suite to rival Gemini Enterprise by connecting cross-app research and automation.
xAI
Released Imagine v0.9 video model with upgraded motion, visuals, and audio.
Meta
Reached 1B monthly users of Meta AI with standalone app.
Miscellaneous
Deloitte Australia will refund part of its AU$440,000 fee after an AI-written government report was found to include fake quotes and references.
Federal prosecutors investigate alleged criminal misuse of AI imagery generation in a California wildfire case, highlighting emerging legal risks in generative AI applications.
📅 AI Events in New Zealand
20 events this week!
This week’s featured event:
AI Forum NZ Launch, AI Governance Refresh, Tues - Auckland. AI Forum NZ is unveiling its refreshed governance site, featuring updated frameworks, expert panels, and practical templates for responsible AI adoption. Perfect for anyone navigating risk, compliance, or AI rollout inside their organisation.
📅 Promote your event with us. Reply to let us know.
💼 AI Roles Around Aotearoa
Picklist of 🌶️ HOT 🌶️ new roles in AI this week.
AI and Analytics Engineer: Auckland
Senior AI & Organic Search Lead: Auckland
💼 Promote your job with us. Reply to let us know.
🤦 ️ AI Fail Of The Week
We all love AI, but it’s certainly far from perfect 🤔 …
Inception line of questioning.
👋 Mike & Erin