Short Answer
Whoever puts the software in front of customers or staff is responsible for it, which in most small businesses means the business itself. The AI tool that wrote the code takes no responsibility for what it produced. A developer sign-off is how you turn that responsibility from something you are carrying blind into something you can stand behind: a named person has read the code, tested it against what it should do, and put their name to a specific version.
Why the Answer Feels Unclear
Before AI coding tools, responsibility followed authorship. If an agency or a freelancer built your booking system, you had a contract with them, they knew how it worked, and when it broke you knew who to ring. The chain was short and every link had a name.
AI breaks that chain at the authorship step. When an owner builds an app by describing it to a tool, a practice now called vibe coding, the code exists but nobody wrote it in the usual sense. The owner directed it without reading it. The tool generated it and has no memory of why it made any particular choice. If a third party built it quickly with AI and has since moved on, there may be no one at all who understands it.
None of that changes who is answerable when something goes wrong. If a customer’s details leak or an automation emails the wrong people, the customer holds the business responsible, and where personal data is involved, data protection law does too. Nobody asks whether a person or a tool wrote the code.
What Responsibility Covers in Practice
Being responsible for software means someone can answer a short list of questions:
- What does it do, and is that what it was meant to do?
- What data does it hold, where, and who can see it?
- What happens when it fails, and would anyone notice?
- Who can fix it, and how quickly?
With AI-built software those answers are often missing because nobody has looked. AI output often works on the cases you tried and fails on the ones you didn’t. Our field note on why AI-built software needs a developer to read it explains where those failures tend to hide. Sometimes the failure is cosmetic. Sometimes it is a page that shows one customer’s records to another. And the tool can be confidently wrong in ways that look identical to being right, the same behaviour behind AI hallucination.
Why Businesses Get It Signed Off
The usual trigger is a change in stakes. A tool built for internal convenience starts taking customer bookings. An automation that drafted emails is switched to sending them. At that point the owner realises they are about to be responsible for something they can’t describe in detail.
A sign-off settles it. A developer reviews the code against a written brief, fixes what fails, and signs a statement naming the version reviewed and the scope it was reviewed against. Responsibility still sits with the business. What has changed is that there is now a written review behind it, and a developer who has read the code.
What to Look For in a Sign-Off
- A written brief first. Nobody can say whether software is right without a description of what right means.
- The code read, not just the app clicked through. Many problems, such as secret keys exposed in the browser, are invisible from the screen.
- A named version and a stated scope. A sign-off that doesn’t say what was checked, and on which version, isn’t worth much.
- A clear line on what happens next. The next AI-generated change is a new version nobody has reviewed. Ask who is accountable after the sign-off, and on what terms.
Common Mistakes
The most common is assuming the tool vendor answers for what the code does once it is running in your business. Their terms put that on you. Close behind is treating “it works when I try it” as evidence that it is safe, when the cases that cause harm are usually the ones nobody tried.
Another mistake comes after a review: carrying on prompting changes straight into the live system. Each of those changes is unreviewed, so within a few weeks the business can be back where it started.
If nobody currently understands the code at all, the problem is closer to a handover than a review, and our article on taking over an existing software project covers what that involves.
How We Approach This
Our AI Sign-Off service is built around this question. We review against a written brief, fix what fails, sign off the version we reviewed, and take on ongoing accountability for the code if you keep us on a support retainer.
A First Step You Can Take Yourself
Before involving anyone, you can check a good deal on your own. Our guide on how to get a vibe-coded app ready for real users lists the checks an owner can run in an afternoon. More answers on related topics are in the Knowledge Center.
Disclaimer: The information provided in this article is for general guidance only and does not override or replace any terms in your contract. While we aim to offer helpful insights through our Knowledge Center, the accuracy of content in this section is not guaranteed.
Written by
Alex
CEO
I’m a software developer and CEO of Digital Royalty, helping growing teams scale their SaaS platforms without losing quality, visibility, or control. I focus on building structured, maintainable systems with clear processes, reporting, and accountability. With over a decade of experience across agency and in-house roles, I specialise in delivering long-term, scalable solutions that support complex, evolving products.