Most indie products shipped in the last couple of years have a meaningful amount of AI-generated code in them. That's not a confession, it's just how software gets built now. But if the product ever matters enough to sell, raise money against, or hand off to an incubator, the question of who actually owns that code stops being academic.
Why the UK's answer is unusual
Most countries, including the US, require a human author for copyright to exist at all. The US Copyright Office has repeatedly refused to register works with no human authorship, meaning fully AI-generated output can end up with no copyright protection whatsoever there.
The UK is different. The Copyright, Designs and Patents Act 1988 has a specific provision for "computer-generated works", ones created by a computer in circumstances where there's no human author, and assigns authorship to "the person by whom the arrangements necessary for the creation of the work are undertaken". In plain terms: if a work genuinely has no human author, UK law still tries to find an owner, usually the person who set the process in motion, rather than leaving it unprotected.
Why this matters less than it sounds for most indie code
That provision is about work with no human author at all. Most AI-assisted coding isn't that. You're prompting, reviewing, editing, rejecting suggestions, restructuring, and directing the overall architecture, that's meaningful human authorship layered on top of AI output, and ordinary copyright rules (you own what you create) apply the way they always have. The genuinely unsettled territory is code that's close to fully autonomous, minimal human direction, minimal edit, effectively "generate an app and ship it as-is".
For a typical vibe-coded side project, most of your codebase almost certainly counts as your own authored work, assembled and directed by you, even where individual functions were drafted by a model. The exposure is at the margins, not the whole codebase.
The part that actually needs checking: your tools' terms of service
Separately from copyright law, every AI coding tool has its own terms governing what you're allowed to do with its output, and these differ and change. Some assign you full rights to use the output commercially with no restriction. Others have more nuanced language around training data, liability, or attribution. None of this is optional reading if the product is going to be sold or licensed, check the current terms of every AI tool in your build pipeline (not just the main assistant, this includes autocomplete tools, code review bots, anything that generated code you shipped), and keep a record of which version of which terms applied when you used it, because terms do get updated.
What to actually do about it
- Keep a rough record of your own contribution. Not a legal audit, just enough that you could describe, honestly, how much of the architecture, logic and editing was yours versus generated. This matters far more once a deal is on the table than it does today.
- Read your AI tools' current terms of service for anything you plan to monetise, specifically the sections on output ownership and commercial use.
- Don't treat this as settled just because it feels fine right now. This is a genuinely developing area of law, in the UK and everywhere else; today's reasonable assumption may not be tomorrow's.
- Get proper legal advice before it's load-bearing, meaning before a funding round, an acquisition conversation, or handing the product to an incubator that will want a clean IP story.
Where Qeetoto fits
This is precisely the kind of question that turns into a blocker halfway through a deal if nobody looked at it earlier. Part of what we do for developers we work with is get the IP picture clean and defensible before it needs to survive scrutiny, while you keep ownership of what you've built. Get in touch if this is a question mark hanging over your project.
FAQ
Does using GitHub Copilot or Claude mean I don't own my app? No. Directed, edited, human-authored use of an AI coding assistant is ordinary software authorship with a powerful tool, not a transfer of ownership. The genuinely uncertain cases are close to fully autonomous generation with minimal human involvement.
Is this different if I'm outside the UK? Yes, meaningfully. Most other jurisdictions require a human author for copyright to attach at all, so fully AI-generated work can end up unprotected rather than owned by someone else. If your company or your customers are outside the UK, get jurisdiction-specific advice.
Should I stop using AI coding tools until this is settled? No. For the vast majority of AI-assisted (not AI-autonomous) development, ordinary authorship principles apply and this isn't a live risk to your day-to-day work. It becomes relevant specifically when the product's value is about to be tested, a sale, a raise, or major licensing deal.
Do I need to disclose AI usage to investors or acquirers? Increasingly, yes, this is becoming a standard due diligence question. Being able to describe honestly how the codebase was built, and having thought about the IP question before being asked, is better than being caught flat-footed.