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.
Who owns the code, copy or designs your startup creates with AI tools?
The reassuring answer is that AI use doesn't automatically mean your business loses control of its work. But it does mean you need to be more deliberate about your contracts, internal policies and client promises.
AI can help you move faster, especially when you're building product features, drafting marketing content or creating design concepts. The risk is that ownership, confidentiality and accuracy can become murky if your documents still assume every deliverable was created by a person from scratch.
Let's look at how to turn that uncertainty into practical protections for your startup.
Can AI-Generated Content Be Protected By Copyright In New Zealand?
It depends, because New Zealand copyright law wasn't written with modern generative AI tools in mind.
The main law is the Copyright Act 1994. Copyright can protect original works such as software code, written content, images, music and designs. In a normal startup workflow, copyright usually sits with the human author first, unless there's an employment rule or contract that transfers it to your business.
AI makes this harder because some output may have little or no human creativity behind it. If your team simply types a basic prompt and copies the output, there may be a real question about whether the result is protected at all.
New Zealand also has a rule for computer-generated works. Broadly, the author can be the person who made the arrangements necessary for creating the work. That may help in some cases, but it won't solve every AI ownership question. The answer can depend on:
- how detailed your prompt was
- how much your team edited, selected or arranged the output
- whether the AI tool's terms let you use the output commercially
- whether the output is too similar to someone else's work
- whether your customer contract says you must create original work yourself
So the practical point is this: don't rely on copyright law alone. Treat AI-generated content as a contract and process issue as well as an IP issue.
Who Owns The Output From The AI Tool?
Your startup may own or control the output, but only if the AI tool's terms and your own contracts line up properly.
Most AI platforms have terms that deal with input, output and licence rights. Some say you own the output. Others give you a broad licence to use it. Some reserve rights to use your prompts and uploads to improve the tool, unless you choose particular privacy settings or a paid enterprise plan.
Before your team uses an AI tool for client work or product development, check:
- Commercial use: are you allowed to use the output in paid products, ads or client deliverables?
- Training use: can the tool provider use your prompts, files or customer data to train its models?
- Exclusivity: could another user receive a similar output?
- Indemnities: does the provider back you if a third party claims the output infringes their IP?
- Restrictions: are there limits on regulated industries, biometric data, scraping, profiling or automated decisions?
Then check your own chain of ownership. If employees are creating work in the course of employment, your business will often own that work. Contractors are different. You should have a written contractor agreement that assigns IP to your company, covers AI use and deals with confidentiality.
From 2026, the Employment Relations Amendment Act 2026 contractor gateway test is also relevant when you're engaging people as contractors in New Zealand. It's mainly about worker status, not copyright, but the bigger point is that your contractor setup should match reality. A good contractor agreement won't fix a misclassified relationship, and it also won't help if it forgets to transfer AI-assisted work to your business.
What Should You Tell Customers About AI Use?
You don't always need to tell customers every time you use AI, but you should be transparent where AI use affects risk, quality, confidentiality or the promises in your contract.
Your customer terms of use should be clear about how you use AI tools in your services. This matters for both trust and legal compliance.
Under the Fair Trading Act 1986, your marketing and contract statements must not be misleading or deceptive. If you advertise custom human-made illustrations but most of the work is generated by AI, that could be a problem. If you say your software has been manually checked for security but you relied on AI output without review, that could also create risk.
The Consumer Guarantees Act 1993 may also matter if you're supplying services to consumers, or to businesses where the Act hasn't been properly excluded for business transactions. Your services need to match what you promised and be carried out with reasonable care and skill.
A Simple Example
Say Aroha runs a Wellington SaaS startup. Her team uses an AI coding assistant to build a reporting dashboard for a client under a $15,000 fixed-fee development agreement.
The client contract says all code will be original, secure and assigned to the client on payment. The AI tool terms allow commercial use, but don't guarantee that the output is unique. Aroha's developer uses the AI-generated code, edits it and runs testing before delivery.
A safer contract position, including any AI output disclaimer, would be:
- Aroha can use AI tools as part of her development process
- her team must review and test AI-assisted work before delivery
- the client receives the agreed licence or assignment to the final deliverable
- Aroha doesn't promise that every line of code was created without AI assistance
- confidential client data won't be entered into public AI tools
That keeps the commercial deal honest and reduces the chance of an argument later.
What Should Your Team's AI Policy Say?
Your AI policy should tell your team which tools they can use, what they can upload and when human review is required.
This doesn't have to be a 30-page manual. For most startups, a practical AI acceptable use policy is enough. It should be written so your team can actually follow it during a busy week.
A policy is especially useful if staff are using free or personal AI accounts. Without clear rules, someone might paste source code, customer data, financial forecasts or investor materials into a tool without checking where that information goes.
The Privacy Act 2020 is a key part of this. If your team enters personal information into an AI tool, you need to think about collection, use, disclosure, storage, overseas access and security. If there's a privacy breach that has caused or is likely to cause serious harm, you may need to notify the Privacy Commissioner and affected people as soon as practicable.
AI Acceptable Use Policy Outline
Here is a practical outline you can use as a starting point. A lawyer should tailor it before you use it, because the right wording depends on your tools, customers and risk level.
- Purpose: explain why your business uses AI and what the policy is trying to protect.
- Approved Tools: list the AI tools your team can use and who can approve new ones.
- Prohibited Inputs: ban confidential information, personal information, client files, passwords, source code or trade secrets unless approved.
- Permitted Uses: describe low-risk uses such as brainstorming, summaries, draft outlines or internal admin.
- Restricted Uses: require approval for client deliverables, code, legal, financial, HR, advertising or regulated content.
- Human Review: set out who checks accuracy, security, bias, plagiarism and brand fit before anything is published or delivered.
- Record Keeping: explain when prompts, outputs, edits and approvals should be saved.
- Customer Disclosure: say when your team must tell a customer that AI was used.
- Privacy And Security: link the policy to your privacy processes and data breach response plan.
- Breach Process: explain what staff should do if they accidentally upload the wrong information.
What Contracts And Warranties Need Updating?
Your customer terms, supplier agreements, employment contracts and contractor agreements may all need AI-specific wording.
The goal isn't to make scary legal documents. It's to make sure your promises match how your startup actually works.
Customer Terms
Your customer terms should say whether you may use AI tools to help provide the services. They should also deal with ownership of the final deliverables, limits on warranties and the customer's responsibilities.
For example, if a customer gives you content, brand assets or datasets to use with AI, your terms should confirm they have the right to provide that material. If your service depends on customer instructions, your terms should say you aren't responsible for problems caused by incomplete or inaccurate inputs.
Employment And Contractor Agreements
Your employment agreements and employee AI use policy should cover confidentiality, IP created during employment and acceptable technology use. Your contractor agreements should go further and include a clear IP assignment to your company.
If contractors can use AI tools, say whether they need your approval, what information they can't upload and whether they must disclose AI-assisted work. This helps you maintain a clean chain of title if you later raise capital, license software or bring on enterprise customers.
Supplier And AI Tool Reviews
If your startup relies on an AI vendor, keep a simple record of your supplier review. Note the tool name, plan type, key terms, privacy settings, data location if available and who approved it.
This record can be useful if a customer asks how you manage AI risk. It also helps your team avoid quietly adopting tools that don't fit your privacy or confidentiality obligations.
A Practical AI IP Checklist For Your Startup
You can reduce most AI ownership risk by getting your workflow and documents aligned early.
Start with the places where AI is already being used. You may find it in product development, customer support, marketing, sales proposals, design, data analysis or internal operations.
Use this checklist as a practical first pass:
- Map Current AI Use: ask your team which tools they use, for what tasks and whether they use personal or business accounts.
- Classify The Risk: separate low-risk internal drafting from higher-risk client deliverables, code, personal information and confidential data.
- Check Tool Terms: confirm commercial use rights, training settings, confidentiality, IP wording and restrictions.
- Update Customer Documents: align your service agreement, SaaS terms, website terms or client proposals with your AI workflow.
- Update Team Documents: add AI rules to your employment agreements, contractor agreements, staff handbook or acceptable use policy.
- Set Review Rules: decide when a human must check accuracy, originality, privacy, security and compliance before output is used.
- Keep Evidence: save material prompts, drafts, edits, approvals and evidence of use for higher-risk projects.
- Review Privacy: check whether personal information is going into AI tools and whether your privacy policy reflects what actually happens.
If something has already gone wrong, act in order. Stop using the output, preserve the prompts and drafts, check the AI tool terms, work out whether personal or confidential information was shared, and get tailored help if a client contract, privacy issue or IP claim is involved.
Templates can be useful for spotting issues, but they often miss the details that matter most for AI use, such as the exact tool settings, the role of human review and the warranties you're comfortable giving.
Key Takeaways
- AI tool IP ownership in NZ isn't just a copyright question. Your contracts, tool terms and internal process all matter.
- AI-generated content may not always have clear copyright protection, especially where there's little human input or originality.
- Your customer terms should be honest about AI use and avoid warranties that don't match your real workflow.
- An AI acceptable use policy helps your team know what they can upload, which tools they can use and when human review is needed.
- Contractor and employment documents should cover AI-assisted work, confidentiality and IP ownership from the start.
- Privacy risks can arise quickly if personal information or confidential client data is pasted into AI tools.
If you'd like help with your AI acceptable use policy, customer terms, contractor IP clauses or privacy wording, you can reach us at 0800 002 184 or team@sprintlaw.co.nz for a free, no-obligations chat.
Protect your brand
What intellectual property should you protect?
If a name, logo, design or other creative work matters to the business, check who owns it, what permissions you need and whether clearance or registration is appropriate.








