Alex is a co-founder of Sprintlaw. Alex previously worked at a top-tier firm as a lawyer specialising in technology and media contracts, and founded a digital agency which he sold in 2015.
"Do we need to change our privacy policy just because we've added an AI chatbot?" In most cases, yes - at least if the chatbot collects names, contact details, order history, account information, support requests or anything else that can identify a person.
The good news is that this doesn't have to be complicated. A clear privacy policy and a few practical supplier terms can go a long way towards meeting your obligations under the Privacy Act 2020 and keeping customer trust intact.
The main thing is to be upfront about what the chatbot collects, where the data goes, who reviews it, and how your business uses it. If the chatbot is powered by a third party AI tool, you'll also want your vendor contract to match what your privacy policy promises.
Do I Need A Privacy Policy For An AI Chatbot?
Yes, if your AI chatbot collects personal information, your business should have a privacy policy that explains how that information is handled.
Under the Privacy Act 2020, personal information is any information about an identifiable individual. That can include obvious details like a customer's name, phone number and email address, but it can also include less obvious details such as:
- chat transcripts that mention a customer's account, order, booking or complaint
- IP addresses, device details or location information that can be linked to a person
- customer IDs, loyalty numbers or support ticket numbers
- voice recordings or uploaded images, if your chatbot accepts them
- free text messages where customers type personal details into the chat box
If your chatbot only answers general questions and doesn't collect anything identifiable, your privacy risk is lower. But many chatbots still collect analytics, conversation logs or technical data in the background, so it's worth checking how the tool actually works before assuming nothing personal is being collected.
The Privacy Act is built around 13 Information Privacy Principles, so your privacy policy should reflect them. For chatbots, the ones that usually matter most are collection, notice, use, storage, security, access, correction and overseas disclosure.
A privacy policy isn't just a website footer document. It should match your real data flows. If your chatbot sends customer conversations to an overseas AI provider, stores transcripts for training, or lets your support team review chats later, your policy should say so in plain English.
What Should Your Chatbot Collection Notice Say?
Your chatbot collection notice should tell customers what you're collecting, why you're collecting it, who may receive it, and what happens if they choose not to provide it.
Information Privacy Principle 3 under the Privacy Act 2020 requires you to take reasonable steps to make people aware of certain details when you collect personal information from them. For an AI chatbot, this usually means you need more than a generic privacy policy hidden several clicks away.
A practical approach is to use a short notice near the chatbot, with a fuller privacy policy linked from that notice. For example, your chatbot pop-up might say that conversations may be recorded and reviewed to answer queries, improve your service and manage customer accounts.
A Simple Chatbot Notice Outline
Your notice should be tailored to your setup, but it commonly covers:
- What the chatbot is: tell customers they're interacting with an AI chatbot or automated assistant.
- What not to share: ask customers not to enter sensitive details unless needed for their request.
- What you collect: mention chat messages, contact details, account details and technical information where relevant.
- Why you collect it: explain that you use it to respond to enquiries, provide support, improve services and keep records.
- Who may receive it: name the types of recipients, such as chatbot providers, hosting providers and your support team.
- Overseas processing: say if information may be stored or processed outside New Zealand.
- Human review: explain if staff or contractors may read chat transcripts.
- Privacy contact: tell customers how to ask for access or correction.
Say Hana runs an online skincare store in Wellington and adds a chatbot to answer product questions. Customers sometimes type in their skin conditions, medication details and order numbers. Hana's short chatbot notice tells customers not to enter medical details unless necessary, explains that chats may be reviewed by the customer support team, and links to a privacy policy that covers her AI vendor. That's much better than saying only "we respect your privacy".
How Should You Deal With Customer Data In Chatbots?
You should collect only what your business genuinely needs, limit who can access it, and set clear data retention rules for how long chat transcripts are kept.
AI chatbots can invite customers to overshare. Someone might type their credit card number, health issue, home address, employment situation or a complaint about another person, even if you didn't ask for it. Once that information sits in your systems, you need to manage it properly.
New Zealand privacy law doesn't create one broad category called "sensitive information" in the same way some overseas privacy laws do. Even so, some types of personal information clearly carry higher risk, such as health information, financial details, children's information, biometric information and information about vulnerable customers. Treat those categories with extra care.
Practical Controls That Help
For most small businesses, the following controls are a sensible starting point:
- configure the chatbot so it doesn't ask for unnecessary personal details
- use warning text such as "please don't enter credit card details or health information in this chat"
- mask or delete payment details and other high-risk information from transcripts where possible
- restrict transcript access to staff who need it for support or quality checks
- set a retention period, such as keeping routine support chats only for as long as needed for customer service and record keeping
- turn off AI model training using your customer data unless you have checked the privacy position and your notices cover it
- keep records of the chatbot provider, data locations, retention settings and information security settings
Customers also have rights to ask for access to and correction of their personal information. Under the Privacy Act, agencies generally need to respond to access and correction requests within 20 working days. If your chatbot stores transcripts, make sure you know how to search, export, correct or delete them when needed.
Can Your AI Chatbot Store Data Overseas?
Yes, your chatbot can store or process customer data overseas, but you need to check the Privacy Act rules and be upfront with customers about it.
Many AI chatbot tools use overseas hosting, global support teams or cloud infrastructure. That doesn't automatically make them unlawful for New Zealand businesses. The key issue is whether you're disclosing personal information to an overseas person or organisation, and whether you can meet Information Privacy Principle 12.
IPP12 generally requires you to have reasonable grounds to believe the overseas recipient is subject to privacy protections that are comparable to the Privacy Act, unless another exception applies. In practice, this means you should look closely at your vendor's privacy and data processing terms, not just the product demo.
Useful questions to ask your provider include:
- Where are chat transcripts, metadata and backups stored?
- Which countries can support staff access the data from?
- Does the provider use your customer conversations to train AI models?
- Can you opt out of model training or product improvement use?
- What security measures protect the data, such as encryption and access controls?
- How quickly will the provider notify you of a data breach?
- Can the provider delete or return data when you stop using the service?
If you can't get clear answers, that's a commercial risk as well as a privacy risk. You may find it hard to answer customer questions, deal with an access request, or respond quickly if something goes wrong.
Your privacy policy should also avoid overpromising. Don't say "we store all personal information in New Zealand" if your chatbot vendor stores backups in another country. A more accurate statement might say that your service providers may store or process personal information outside New Zealand, and that you take steps to choose providers with suitable privacy and security protections.
What Should Your AI Chatbot Privacy Policy Include?
Your AI chatbot privacy policy should describe your actual data practices, not just copy a generic privacy template.
Templates can be a useful starting point, but they often miss AI-specific issues like human review, prompt logs, model training, overseas hosting and vendor access. A lawyer should tailor the policy before you use it, especially if your chatbot deals with health, finance, children, employment, complaints or high-volume customer support.
Fill-In Outline For Your Chatbot Privacy Policy
- Who You Are: include your legal entity name, trading name, NZBN if relevant, and privacy contact details.
- What The Chatbot Does: explain that you use an AI chatbot or automated support tool on your website or customer support channels.
- Information You Collect: list chat messages, contact details, account details, order details, technical data and any uploaded files.
- Information Customers Should Avoid Sharing: tell customers not to provide payment card details, passwords or unnecessary sensitive details in the chat.
- Why You Use The Information: cover customer support, responding to enquiries, managing orders, improving services, fraud prevention and record keeping.
- AI Processing And Training: state whether conversations may be used to improve the chatbot or train models, and whether customers can opt out.
- Human Review: explain when your team, contractors or provider support staff may review transcripts.
- Who You Share Information With: describe chatbot providers, cloud hosts, CRM tools, payment providers and other service providers.
- Overseas Storage Or Access: say whether information may be processed outside New Zealand and how you manage that risk.
- Security And Retention: explain your main safeguards and how long you keep chatbot records.
- Access And Correction Rights: explain how customers can request access to or correction of their personal information.
- Privacy Breaches: explain that you'll assess and notify notifiable privacy breaches where required by the Privacy Act 2020.
You should also make sure your privacy policy lines up with your website terms, chatbot terms of use and internal staff procedures. If one document says chats are deleted after 30 days but your vendor keeps them for 12 months, you'll need to fix that mismatch.
What Vendor Terms Should You Check Before Using A Chatbot?
You should check that your chatbot vendor contract gives you enough control over customer data and doesn't undermine your privacy promises.
Most AI chatbot tools come with standard online terms. Those terms can be fine for a low-risk tool, but they may not be enough if you're sending customer account data, support histories or high-risk personal information through the chatbot.
Before you launch, and as part of your AI governance process, look for clauses that cover:
- Role Of The Provider: whether the provider acts as your service provider, processor or independent controller of the data.
- Permitted Use: the provider should only use customer data to provide the service, unless you've clearly agreed otherwise.
- AI Training: whether customer conversations can be used to train or improve AI models.
- Subcontractors: whether the provider can use other vendors and where those vendors are located.
- Security Standards: access controls, encryption, logging, backups and staff confidentiality obligations.
- Breach Notification: how quickly the provider must tell you about a security or privacy incident.
- Access And Correction Support: whether the provider will help you respond to Privacy Act requests within the required timeframe.
- Deletion And Return: what happens to transcripts and backups when you cancel the service.
- Audit Or Assurance: whether the provider offers security reports, certifications or other comfort around its controls.
If a provider's terms say it can use all inputs and outputs for its own product development, but your privacy policy says customer data is used only for support, you have a problem. Either your use of the tool needs to change, or your documents and customer notices need to be updated accurately.
If something goes wrong, act in order. Preserve the chat logs, identify what information was involved, ask the vendor for incident details, work out whether serious harm is likely, and keep a written record of your decision. Notifiable privacy breaches must be reported to the Office of the Privacy Commissioner and affected individuals as soon as practicable where the legal test is met.
Key Takeaways
- If your AI chatbot collects personal information, your privacy policy should explain how that data is collected, used, stored, shared and protected.
- Your chatbot should have a clear collection notice, especially if conversations are recorded, reviewed by humans or sent to a third party provider.
- Customer data in chatbots can include more than names and emails. Chat transcripts, account details, IP addresses and uploaded files can all be personal information.
- Overseas storage is common with AI tools, but you need to check Privacy Act 2020 requirements and be honest about data locations in your privacy documents.
- Your vendor terms should cover AI training, subcontractors, security, breach notification, access requests and deletion when the service ends.
- Generic templates often miss AI-specific issues, so it's worth tailoring your privacy policy and supplier terms to your real chatbot setup.
If you'd like help with your AI chatbot privacy policy, customer collection notices, website privacy documents or supplier data processing terms, you can reach us at 0800 002 184 or team@sprintlaw.co.nz for a free, no-obligations chat.
Drafted by Taylor, Sprintlaw's AI, and overseen by Alex Solo. How we write
Get your customer-facing terms right
What should your privacy and online terms cover?
If you collect customer data, sell online or run marketing campaigns, your public terms and privacy documents should match the real customer journey.








