I was taken aback recently when a response from an AI tool became unexpectedly personal. The response started with “As a board chair or trustee reviewing this situation…”
My immediate thought was: You’re a what? A board chair?
It was placing human qualities and lived experience onto something that has neither. AI is not a board chair or trustee. It does not carry the responsibilities that come with those roles, and we shouldn’t confuse its ability to sound human without actually being human.
If you use AI enough, you start noticing some of its other human-like responses. Question something it got wrong, and it apologises like it’s been caught out, with a “my bad, sorry” and lots of smiley faces. I’ve also noticed its responses and quality of information change when the technology is updated. Equally, what we feed the system matters, and so does holding it to account.
There is so much noise in the AI conversation right now. Everyone has an idea, a view, an opinion or a new tool to recommend. The challenge is distilling what is genuine, useful and factual, because the lines are increasingly being blurred.
Recently, in less than five minutes, I uploaded an image of myself and created a two-minute presentation featuring what appeared to be me speaking at a Ted Talk-style event, delivering a script I had never read, and looking very professional. Sounds pretty good, right?
AI might be clever, but it can also be unnerving. Images, video, algorithms, information and increasingly sophisticated AI-generated content are making it harder to distinguish between what is real, what was generated, and where one starts and the other stops.
Using an AI tool, or being used?
AI does not bring lived governance experience to the board table. It does not know our communities, understand the relationships behind our decisions, or carry responsibility for what happens next. Yet a polished, confident and personal-sounding answer can make that surprisingly easy to forget.
The question is not whether we should embrace AI or reject it. The question is whether we understand what we are using, why we are using it and what conditions we put around it before we give it our information, our trust, or influence over our decisions.
Being cautious when those boundaries are unclear is not falling behind. This is good governance.
The potential is real. AI can reduce administration, help us work through information faster and potentially give us more time for our kaupapa. But, sometimes the technology itself reminds us why safeguards matter.
In July 2026, during internal cybersecurity evaluations, OpenAI models circumvented controls intended to isolate them from the internet. They communicated through unauthorised channels, exploited vulnerabilities, gained internet access and broke into Hugging Face’s systems. The incident was subsequently examined by OpenAI and independent researchers.
This was a specialised evaluation environment, not an everyday chatbot simply ignoring a user’s instructions, but it provides an important governance lesson: instructions alone are not safeguards. Conditions need to work in practice, with monitoring, accountability and the ability to intervene when something goes wrong.
Our own community sector is asking for that care.
Community Governance Aotearoa’s 2026 Sector & Insights Survey received 201 responses. 55% were concerned about using AI tools responsibly. 39% reported low or no confidence in making decisions about data, privacy or AI, and only 8% were very confident. Just 38% reported having some form of AI policy or guidelines.
I don’t read those numbers as communities resisting change.
I read them as communities asking how we do this well.
Upcoming conversations:
BoardPro + Community Governance Aotearoa: Governing with Impact Roadshow
Auckland – Friday 30 October 2026 – Crowne Plaza, 128 Albert Street (SOLD OUT)
Wellington – Monday 2 November 2026, Rydges, 75 Featherston Street
Christchurch – Wednesday 4 November 2026, Rydges, 30 Latimer Square
Roadshow information and tickets
Data Talks: Data Governance – From Local Practice to Global Perspectives
Wednesday 11 November 2026, 12-1pm, online and free to attend. The session includes guest speaker Anthony Wong, exploring global practice, Indigenous perspectives and community-led approaches. Register for the November Data Talk
2026 Sector & Insights Survey
