Woman presenting on AI safety
  • Human Resources
  • Health and Safety
  • HR and Health and Safety Consulting

AI in Safety: what's happening now in NZ

Joe Lill
Joe Lill
3 min read
Share

All industries are grappling with how best to integrate AI into their operations and safety is no different. Safety teams have spent the past year working out what AI is good for. The answer so far: pattern spotting, data review and video analysis at scale and beyond what site teams could previously manage.

Reviewing years of incident reports by hand to find a recurring near-miss might take a person a week. An AI tool can do it in minutes.

Reducing or translating safety information is now accessible and quickly done to a reasonable standard. AI-integrated cameras can be trained to identify workers in dangerous positions during loading and unloading or missing PPE. The speed and consistency of these reports drive real change in the conversations and accountability of workers.

Areas where judgement and decision making are required are still a step too far. The need for expert safety support is still needed to ensure outputs are at the required standard. The conversation has shifted, however, from what data entry or document preparation costs, to where are the opportunities to add more value with your teams or leaders.

Where to start

Begin with information, not decisions. Work out your data sources, for example telematics, incident reports, leave records and start experimenting. Review your hazard and risk register to identify gaps or duplications. Review policies or procedures against best practice to see what controls are missing or not present. Turn a dense guidance document into something a site team can actually read. Keep the first uses low-stakes and reversible. Get a feel for where it's helpful.

A risk you need to be aware of is the confidence of AI. The speed and presentation of AI make it seem right even when it is entirely wrong. Fabricated or made-up references or standards happen. AI looks for an answer and will give you one even if it has to make it up. Here are some ways to manage the risk:

  • Judgement first. treat AI output as a draft from a keen but unqualified person: useful, fast, but needing a competent person to check it. That person needs to be familiar with the subject matter and able to check it against legislation and/or current WorkSafe guidance.

  • Human sign-off. A risk assessment, a permit-to-work, a critical control, an incident classification: these get a trained and competent person's eyes before they're actioned. AI-spotted issues are still reviewed before a disciplinary letter is sent or a conversation is held. AI drafts - but people still hold decision-making power.

  • Ground the prompt in your own material. Instead of asking an AI tool what the rules are, give it your actual policy or your incident data (taking care with privacy) and ask it to work within that. The better the context you provide, the less room there is for the tool to invent an answer.

The workplaces getting value from this aren't the ones treating AI as an oracle. They're the ones using it as a very fast assistant, with a competent person still holding the pen especially on anything that could hurt someone.

How we can help

Working across a wide range of industries, we are seeing AI used in creative and interesting ways. Its integration into safety software is increasingly common, and making sure it's set up properly and delivering value is an area we can help with.

We're increasingly asked to give a final review or provide feedback on processes and documents clients have started with AI. It speeds up the groundwork and lets us put our time into the parts that need a competent person's judgement, so you get the benefit of both the speed and the expertise.

Disclaimer

Findex NZ Limited trading as Findex.

The views and opinions expressed in this article are those of the author/s and do not necessarily reflect the thought or position of Findex. This article contains general information and does not constitute legal or taxation advice. If you need legal or taxation advice, we recommend you speak to a qualified adviser.

November 2026

Joe Lill
Author: Joe Lill | Associate Partner

Joe brings over a decade of experience as a health and safety lawyer to his role at Findex. With a passion for creating safer work environments, he leverages his expertise to assist clients enhance their health and safety systems through design, audit, and comprehensive investigations. Joe works to empower clients with the knowledge and tools they need to manage worker health and safety with confidence.