272 active AI use cases in NZ's Public Service

PLUS, introducing AI On The Couch

Kia ora! Welcome to New Zealand’s weekly roundup of AI news, events, jobs and education.

Two announcements from this weekend:

  1. We wrapped up the AI Forum’s NZ AI Hackathon: 48 hours jamming on how AI can help tackle the UN’s Sustainable Development Goals (with NZ only on track to hit 16% of our goals by 2030 🤯). Stoked to share our idea picked up first place 🥇 and a spot at regionals. Way to go team.

  2. We’ve just kicked off AI On The Couch, born from Erin’s endless curiosity about AI prompts / tools / concepts, which sparked the idea 👉 she asks, I try answer, all in 5-10 min episodes. First up: ChatGPT’s new Agent Mode.

    1. 🎥 YouTube 

    2. 🎧 Spotify

This week’s highlights

📉 Kiwi retailers fall behind on AI
🚨 Privacy chief warns of AI risks
⚖️ AI evidence rejected in mock trial
🧠 NZ copyright law backs AI creators
🛠️ Huckleberry slashes feedback time
🌱 We need AI gardeners, not carpenters
⚖️ Court rejects 'black-box' AI in mock trial

Happy reading and listening ✌️

Did someone forward you this? Sign up!

HEADLINE STORY

📊 AI Transforms Kiwi Public Service Productivity

📣 Word On The Street: 70 government agencies now using AI tools, up from 44 last year, with Microsoft Copilot leading the charge.

🔍 Zooming In:

  • 272 AI use cases reported across public service, up from 15 operational cases in 2024

  • 55 AI systems now fully deployed, helping automate admin tasks and improve efficiency

  • Microsoft Copilot handles half the workload, with custom ChatGPT tools making up 25%

🏘️ Our Take: The jump in AI system use reveals something profound: government is moving from cautious experimentation to confident deployment. When a public servant can transcribe calls in 2 minutes instead of 10, that's not just time saved; it's a fundamental shift toward responsive governance. The real win here is that agencies are sharing AI tools rather than building silos, creating a collaborative ecosystem that could transform everything from passport processing to benefit applications. The focus on "supporting functions" rather than customer-facing services shows smart prioritisation, get the back-office right first, then transform citizen experiences.

Microsoft's dominance shows we're leveraging existing infrastructure smartly, but the 25% custom solutions suggest our public service is innovating. As with any risk of custom solutions, maintenance and management can limit scalability. Skills shortages remain the biggest hurdle, which means investment in training could unlock exponential improvements in how government serves citizens.

🐝 Save hours of work starting today.

Try Autohive free and build AI agents in minutes that tackle your most repetitive tasks - secure, private, and ready when you are.

🏛️ Government, Policy & Legal

Kiwi courts reject AI evidence in mock drug trial, raising tech liability questions. High Court Justice Greg Blanchard KC ruled a fictional search warrant based on "TraceLex" AI analysis was likely invalid due to black-box algorithms and lack of peer review. The audience overwhelmingly agreed AI evidence should be inadmissible without supporting proof.
3-min read.

  • Our take: Signals major compliance headaches ahead for any business developing AI tools for law enforcement or legal discovery. The "black box" problem could render expensive AI systems legally worthless if algorithms can't be scrutinised or validated. Companies building AI for sensitive applications need transparent, auditable systems from day one, not afterthoughts. The argument for commercial sensitivity won't fly when liberty is at stake.

Privacy Commissioner warns Government's "light touch" AI approach creates regulatory gaps. Michael Webster flagged concerns about Shane Reti's hands-off strategy, citing Australia's RoboDebt and UK's Horizon scandals where thousands were harmed by inadequate risk management. Meanwhile, Accenture-Microsoft report promises $75b economic boost by 2038.
4-min read.

  • Our take: This sets up a classic regulatory dilemma: move fast and break things, or go slow and miss opportunities. The Privacy Commissioner's concerns are valid given overseas disasters, but businesses need clarity to invest confidently. The real issue is that our Privacy Act wasn't designed for AI's data-hungry nature, creating compliance uncertainty that could stifle innovation.

Kiwi copyright law accidentally backs AI-generated images whilst US blocks them. University of Auckland research shows GenAI outputs can gain 50-year copyright protection as "computer-generated works" if they meet originality tests.
21 pages.

  • Our take: This puts Kiwi businesses in a surprisingly strong position. If you create AI-generated content here, you can legally own and protect it, something American competitors simply can't do. Think marketing materials, product designs, or digital content created with AI tools. The catch is the protection isn't as strong as traditional copyright, and you'll need to clearly mark everything as AI-generated. The window won't last forever though, either other countries will catch up or our laws will change.

💼 Business & Industry

Kiwi retailers lag behind Asia in AI adoption race. New Shopify research reveals 17% of ANZ merchants won't touch AI in 2025. Nearly three times higher than other APAC markets at 6%. Retail employs hundreds of thousands across both countries.
2-min read.

  • Our take: This hesitation could cost us serious competitive ground when Asian retailers are racing ahead with AI-powered personalisation and fraud detection. The timing is brutal, retail margins are already under pressure, and this productivity gap will only widen. We need targeted digital infrastructure investment and AI capability programmes, while balancing a challenging economic outlook. The economic stakes are massive given retail's employment footprint.

Kiwi startup uses AI to solve workplace feedback nightmares. Huckleberry slashes 360-degree feedback time by 90% using voice-based AI analysis instead of lengthy surveys.
3-min read.

  • Our take: This tackles a genuine business pain point that's only gotten worse with remote work: most companies avoid comprehensive feedback because traditional 360 reviews can consume hundreds of management hours with questionable ROI. Also, imagine hiring based on verified peer feedback rather than polished interview performances. The real test will be whether AI can capture nuanced interpersonal dynamics or just surface-level pleasantries.

🎙️ The AI Corner podcast

This week’s guest is Dan Xu, co-founder and CEO of Element X.

Dan shares how he’s been building practical AI solutions for over a decade, long before the ChatGPT hype. From his roots in bioengineering to helping enterprises adopt AI safely and meaningfully, Dan explains why real-world value, human-centred design, and smart implementation matter more than buzzwords.

🎧 Listen on Spotify or YouTube.

Subscribe on Spotify and YouTube to be notified of new episodes.

📚️ Levelling Up With AI

Practical AI for Everyday Work

1️⃣ Nadia Ellis’ AI tips cut through the hype with 10 dead-simple ways to make AI useful in daily work.
2-min read.

2️⃣ I’m not a developer, but with Relay.app’s drag-and-drop agents I’m automating LinkedIn feeds, newsletters, and daily summaries. No coding needed.
85-sec demo.

The Speed & Scale of AI

3️⃣ I asked ChatGPT’s Agent Mode to book me a tradie. Five minutes later, my phone rang with a contractor ready to go.
66-sec watch.

4️⃣ NZ’s 300 Spartans. The AI future belongs to ‘Tiny Teams’: small, high-trust squads using AI fluency, automation, and rapid builds to outpace bigger rivals.
4-min read.

Leading in the AI Era

5️⃣ My big bet: The companies that win in AI’s first innings will promote Gen-Z AI natives into decision-making roles years ahead of schedule. Waiting a decade for “paying dues” is a growth killer.
4-min read.

6️⃣ Why NZ businesses need fewer AI Carpenters and more AI Gardeners. An AI advantage won’t come from perfect plans, it’ll come from leaders who plant widely, nurture what works, and rip out what doesn’t.
3-min read.

🌍 The News from Global

US embeds trackers in AI chips to catch illegal exports to China. Enforcement agencies secretly place tracking devices in Dell and Super Micro servers containing Nvidia and AMD chips to monitor diversions from export restrictions.
4-min read.

💊 MIT develops AI-designed antibiotics against superbugs in major medical breakthrough. Two new drugs created through artificial intelligence show promise against antibiotic-resistant bacteria, currently in animal testing phase.
3-min read.

Tech Updates You Should Know

  • OpenAI: Added smart router for dynamic model selection, raising user control concerns; restored GPT-4 and GPT-4o access for Plus users and doubled message caps; announced $12.7B projected revenue for 2025 with 700M weekly users; rolled out GPT-5 across all Microsoft platforms; planning 5-month compute capacity expansion; launched advertising tests, explored government integrations; Sam Altman in talks to invest in brain-computer interface firm Merge Labs; GPT app reached $2B in mobile consumer spend.

  • Google: Fixed erratic behaviour bug in Gemini AI and launched Guided Learning mode for interactive tutoring; introduced memory, privacy settings, and temporary chat features in Gemini; partnered with NASA to build CMO-DA AI medical assistant for space missions; launched "Create" tab in Photos for AI-powered editing; NotebookLM updated with video summaries; unveiled Opal no-code builder and Genie3 immersive tool; received $34.5B unsolicited bid for Chrome from Perplexity amid DOJ antitrust scrutiny.

  • Anthropic: Launched opt-in memory and 1M-token context window; introduced Learning Mode for step-by-step coding assistance; offered Claude access to all three U.S. government branches for $1/year in strategic match to OpenAI; criticized Meta for talent poaching and addressed excessive AI compensation concerns.

  • Microsoft: Released Copilot 3D for image-to-3D model conversion across apps; deployed GPT-5 across Copilot, GitHub, Azure, and VS Code, reaching 100M users; added smart model switching in Copilot; intensifying recruitment of Meta AI researchers; integrating GitHub deeper into AI strategy post-leadership change.

  • xAI: Released Grok 4 globally for free with Auto and Expert modes and image-to-video generation via Grok Imagine; added NSFW "Spicy Mode" for premium X users; faced Grok 4 defeat by GPT-5 in Kaggle chess final; threatened legal action against Apple for App Store favouritism toward ChatGPT; announced lawsuit to block OpenAI’s for-profit transition; criticised for chatbot policy lapses and deepfake risks.

  • Meta: Acquired WaveForms AI for emotional speech synthesis; faced backlash over leaked chatbot policies allowing inappropriate conversations with minors, revised rules post-scrutiny; partnered with Anduril for military-grade VR development.

  • NVIDIA: CEO Jensen Huang spotlighted robotics and simulation as AI’s next frontier; introduced open multilingual speech dataset and Cosmos Reason for robotic agents; agreed to 15% revenue-sharing with Chinese government for export license retention; AI chip shipments now tracked by U.S. government.

  • Apple: Testing App Intents for Siri to control third-party apps via voice commands; integrating GPT-5 into Siri and iOS 26 for enhanced AI experiences; exploring offline-capable AI model partnerships; faces rising regulatory scrutiny in global AI safety debate.

  • Perplexity: Proposed $34.5B unsolicited bid to acquire Google Chrome amid antitrust challenges; launched AI-powered search on Truth Social; criticised for being easily replicable by larger tech companies.

📅 AI Events in New Zealand

21 AI events are happening around New Zealand this week. A bit light in the regions — so if we’ve missed anything, let us know and we’ll add it in.

This week’s featured event:

  • Wellington Before9: Responsible AI, Wellington - Wed: With the government’s first national AI strategy now live, the big question is what “responsible AI” really means in practice for Kiwi businesses and organisations. This panel will explore the opportunities and risks of a light-touch regulatory approach, and how we can shape a uniquely Aotearoa framework for AI adoption.

📅  Promote your event with us. Reply to let us know.

💼 AI Roles Around Aotearoa

Picklist of 🌶️ HOT 🌶️ new roles in AI this week.

💼  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 🤔

Looks the best spot for solar panels.

👋 Mike & Erin