Artificial intelligence has moved from novelty to routine fixture in New Zealand workplaces almost overnight.

Employees draft emails with it, managers use it to prepare performance documentation, and a growing number of staff are even wearing it – literally – in the form of AI-enabled smart glasses.

For employment lawyers, HR teams, and the employees and employers we advise, this shift raises two connected but distinct problems: what happens to legal privilege and confidentiality when sensitive employment material is fed into an AI tool, and what happens to workplace privacy when AI is capable of quietly recording everything around it.

Both issues are unsettled in New Zealand law. Both carry real consequences if they go wrong.

Part 1: Does pasting a “private and confidential” letter into AI put your privilege at risk?

It’s an increasingly common scene in employment disputes: an employee receives a “private and confidential” or “without prejudice” letter from their employer – perhaps a settlement offer, a disciplinary outcome, or restructuring proposal – and, wanting a quick second opinion, pastes the whole document into ChatGPT, Claude, or another AI tool to ask “what does this mean” or “how should I respond.”

It feels harmless. It may not be.

Why confidentiality is the whole ballgame

Legal privilege in New Zealand depends heavily on communications remaining confidential. Uploading sensitive material into a publicly available AI tool can jeopardise that privilege, particularly where the tool’s terms and conditions allow the provider to collect, use, or disclose what is typed into it.

Critically, AI tools do not qualify as a “legal advisor” under the Evidence Act 2006 – so even if an AI chatbot sounds convincingly like a lawyer, communications with it will not attract legal advice privilege in the way a conversation with an actual solicitor would.

This matters enormously in the employment context, where “private and confidential” and “without prejudice” correspondence is a daily feature of restructuring, disciplinary processes, personal grievances, and settlement negotiations.

If that material – or an employee’s own analysis of it – ends up disclosable in a later Employment Relations Authority or Employment Court proceeding because privilege has been undermined, the consequences can be serious.

What the overseas courts have said

New Zealand courts have not yet ruled directly on this issue, but courts in the United States and United Kingdom have started to, and their reasoning offers a preview of the arguments likely to surface here:

• In one US case, a court held that privilege did not attach to a defendant’s exchanges with an AI chatbot about his legal position, because the AI was not a lawyer and the version used was public, with terms allowing the provider to use and potentially disclose the input.
• In two US employment discrimination cases involving self-represented plaintiffs who used AI to help draft legal documents, the courts reached the opposite conclusion – finding privilege was not waived, on the reasoning that an AI programme is a “tool, not a person,” so no third-party disclosure had occurred.
• A UK tribunal, dealing with lawyers who had filed AI-generated submissions containing fabricated case citations, went further: it found that uploading confidential documents into a public AI tool effectively places that information in the public domain, amounting to both a breach of client confidentiality and a waiver of privilege.

The common thread is that outcomes are fact-specific, and depend heavily on which AI tool was used, how public or enterprise-grade it was, and what its terms and conditions say about your data.

The practical risk for employees and employers

For employees, pasting confidential employer correspondence into a public AI tool can:

• Undermine any later claim of privilege over their own communications about the dispute.
• Potentially breach confidentiality obligations in their employment agreement, particularly where the letter is marked “private and confidential” and relates to a live disciplinary, restructuring, or investigation process.
• Expose commercially sensitive employer information (names, remuneration details, other staff members’ personal information) to a third-party AI provider’s servers, which may itself be a separate privacy issue.

For employers, the risk runs the other way: confidential HR material, legal advice, or settlement strategy pasted into AI tools by managers or advisors – without appropriate enterprise safeguards – can similarly waive privilege and undermine an organisation’s position in later litigation.

Until New Zealand law is clearer, the safest course for both sides is to treat privileged or confidential employment material the same way you would treat it before AI existed: don’t share it outside a closed, trusted circle without first getting legal advice on the consequences of doing so.

If AI assistance is wanted, licensed, enterprise-grade tools with contractual confidentiality protections offer materially better protection than free, public consumer tools.

Part 2: The other AI risk employers haven’t clocked yet – smart glasses

While much of the privilege debate has focused on generative AI chatbots, a second and quite different AI-related risk is creeping into New Zealand workplaces: AI-enabled smart glasses, most visibly Meta’s Ray-Ban range, due to launch in New Zealand in September 2026, and now joined by budget alternatives such as Kmart’s $99 Anko camera glasses.

These devices look like ordinary eyewear but contain a camera, microphone, speakers, and an AI assistant capable of recording, transcribing, and analysing whatever the wearer sees and hears – all activated discreetly, with only a small LED light (which can be obscured) indicating that recording is underway.

A regulatory gap, not just an etiquette problem

New Zealand privacy experts have been blunt about the problem this creates.

As University of Canterbury professor Annette Mills has observed, these devices are “essentially a computer on your face,” and the Privacy Act 2020 was not built with this scenario in mind.

The Office of the Privacy Commissioner has confirmed that the Act is largely aimed at regulating organisations, not individuals – meaning an employee or visitor wearing smart glasses in a workplace generally only breaches the Act if their collection, use, or disclosure of what they capture is “highly offensive to a reasonable person.”

That is a high, and highly fact-dependent, threshold. A great deal of everyday recording by individuals simply falls outside the Act’s protections altogether.

That gap has direct implications for workplaces:

• Colleague-to-colleague recording. An employee wearing smart glasses in a meeting, on the shop floor, or in a client interaction could be capturing audio and video of colleagues, clients, or confidential documents without anyone else’s knowledge or consent – and current NZ privacy law may not clearly prohibit it.
• Client and patient-facing roles. In professional services, healthcare, retail, and hospitality settings, staff wearing smart glasses raises acute confidentiality and trust concerns – a client discussing a legal or medical matter has no reliable way of knowing whether they’re being recorded, transcribed, or fed into an AI system for later analysis.
• Evidence in employment disputes. Footage or audio captured by an employee’s smart glasses – whether of a disciplinary meeting, a workplace disagreement, or a colleague’s private conversation – may end up as evidence in a personal grievance. New Zealand courts and the Employment Relations Authority have generally taken a dim view of covert recording between colleagues, particularly where it breaches an employment policy or the trust that is fundamental to the employment relationship, and such recordings have been excluded from evidence in past cases.

What employers should be doing now

Because generic Privacy Act protections may not go far enough, the practical answer for employers lies in workplace policy, not waiting for legislative reform:

1. Update your IT, confidentiality, and code of conduct policies to explicitly address wearable recording devices, not just phones and traditional cameras – smart glasses are easy to overlook because they don’t look like recording equipment.
2. Set clear rules for client and patient-facing areas, meeting rooms, and any space where sensitive personal or commercial information is discussed, including a right to require staff and visitors to remove or disable smart glasses.
3. Address AI chatbot use and wearable AI devices together in a single, practical AI-use policy, since both raise overlapping confidentiality, privacy, and privilege risks.
4. Treat covert recording by staff – whether via smart glasses, phones, or otherwise – as a serious disciplinary matter, consistent with the approach NZ courts have taken to secretly recorded workplace conversations.
5. Be transparent about any monitoring the business itself carries out, since covert employer surveillance of staff remains high-risk and is generally only justifiable in exceptional, well-documented circumstances such as a credible suspicion of serious misconduct.

The common thread

Both issues covered here – AI chatbots undermining legal privilege, and AI-enabled glasses undermining workplace privacy – share the same underlying problem: New Zealand’s legal framework was not designed with these technologies in mind, and is being asked to stretch to cover them using existing principles of confidentiality, privilege, and privacy.

Until Parliament or the courts provide clearer answers, the safest approach for both employers and employees is caution: treat confidential and privileged material with the same care you always have, be alive to the possibility that a colleague’s “glasses” are actually recording equipment, and put clear, specific policies in place now rather than after something has gone wrong.

If you have questions about how these issues might affect a live employment dispute, restructuring process, or workplace policy, it’s worth getting specific legal advice before sensitive material goes anywhere near an AI tool – public or otherwise.

Need support? If you would like assistance with any employment matter relating to AI, please get in touch with our team. Email admin@cmalaw.co.nz or phone 03 211 0153.

Disclaimer: This update provides commentary on employment law, health and safety and immigration topics, it should not be used as a substitute for legal or professional advice for specific situations. Please seek legal advice from your lawyer for any questions specific to your workplace.