Short Answer
No, vibe coding is not bad in itself, but it is risky in a specific set of situations, and most of the arguments about it online miss which ones. Building software by describing it to an AI tool and accepting what it writes is a sensible way to make a prototype or a small internal tool. It becomes a problem when the result starts holding customer data, taking money, sending messages or changing records on its own, because nobody has read the code that now carries those risks. The question for a business owner is less “is vibe coding bad?” and more “what is this particular app doing, and has anyone checked it?”
Why the Question Gets Asked So Much
The term is young. Andrej Karpathy, a founding engineer at OpenAI, used it in February 2025 to describe building an app while being able to “forget that the code even exists”. By November 2025 Collins Dictionary had named “vibe coding” its Word of the Year, pointing to a sharp rise in use over those nine months. Our glossary entry on vibe coding covers the definition.
Search for the question and most of what comes back is written by developers for developers: forum threads and engineering blogs arguing about whether it makes programmers worse at their jobs. That debate matters to developers. If you built a quoting tool over a weekend, or a member of staff built one for you, you want to know whether it is safe to rely on. The rest of this page answers that.
Where Vibe Coding Is Fine
A lot of software a small business needs is low stakes, and for that kind of work vibe coding is a good trade. You get a tool shaped around how you work, in days, for the cost of a subscription.
It is usually fine when:
- the users are you or a few colleagues who would notice a wrong answer and say so
- the data in it is not personal, or is already public
- nothing happens without a person pressing a button
- if it stopped working tomorrow, you could go back to the spreadsheet without much pain
A holiday rota for a team of eight is a good example. If it double-counts someone’s leave, the person whose leave it is will spot it within a day. The cost of a bug is a short conversation. Spending money on a developer review for that would be hard to justify.
Prototypes are the other strong case. Showing a working version to a potential customer or investor is far more convincing than a slide, and if the idea goes nowhere you have lost very little.
Where It Bites
The trouble starts when the same build process is used for something with higher stakes, and nothing about the process changes to match. Four areas cause most of the damage.
Security. AI tools write code that does what you described. They do not reliably add the protections you didn’t describe. Login checks that confirm someone is signed in without checking which records are theirs, secret keys sitting in code the browser can read, and forms that pass whatever is typed straight into a database are all common in AI output, along with admin pages open to anyone who finds the address. None of them show on the screen, so the app looks finished to the person who built it.
Data. Once an app holds names, emails, addresses or anything more sensitive, UK GDPR applies to it in the same way it applies to software a developer wrote. If personal data leaks, the business has to assess it and, where it is likely to put people at risk, report it to the ICO within 72 hours. That is hard to do for an app when nobody can say where its data is stored or who can see it.
Maintenance. Vibe-coded apps are often easy to start and hard to change. Each new prompt adds code on top of the last, and the tool has no memory of why it made earlier choices, so after a few months the app can be held together by fixes on top of fixes. That build-up is a form of technical debt, and it usually shows up as the day a small change breaks something unrelated and nobody can work out why.
Ownership. Some apps live inside the platform they were built on, under one person’s login. When that person leaves, or the subscription lapses, the business can find it has an app it depends on and no clear right or route to the code. Who is answerable when something goes wrong is a separate question, and our article on who is responsible for software built with AI covers it.
Take a trades business that builds a quote calculator with an AI tool. At first the calculator is internal: the owner types in measurements and gets a price. Then it goes on the website so customers can get their own quotes, and later it starts saving their contact details so the office can follow up. Each step is small, and each one seems like an obvious improvement. But the app that was a private calculator is now a public form collecting personal data, and nobody has looked at it since the first weekend.
What About AI Code Review?
A common answer to the risks above is to point a second AI at the code. AI code review tools read a codebase and flag likely problems: exposed keys, unsafe database queries, packages with known flaws, errors that are ignored. They cost little and run every time the code changes, so for anyone building with AI they are worth switching on.
They have two limits. They judge code against patterns of what usually goes wrong, so they cannot tell you whether the app does what your business meant it to do. A discount that applies twice or a booking slot that can be taken by two customers is perfectly valid code; it is only wrong against your intentions. And a tool’s report is a list of comments with no one behind it. A clean report means the tool found nothing it recognised. Our comparison of AI code review tools and a developer sign-off goes through where each is enough.
What to Look For in Your Own App
You can get a fair read on where your app sits without any technical knowledge. Ask:
- Who uses it? Only your team, or customers and the public?
- Does it store anything about real people?
- Can it send messages, take payments or change records without someone approving each one?
- If it went wrong for a week without anyone noticing, what would it cost?
- Could anyone other than the person who built it change it?
If every answer points to “low stakes”, carry on and enjoy the time it saves. If two or three don’t, the app has moved into the territory where vibe coding causes problems, whatever it was when it started. Our guide on getting a vibe-coded app ready for real users walks through the checks an owner can run in an afternoon.
Common Mistakes
The most common mistake is judging an app by whether it works when you try it. You tried the cases you thought of. The cases that cause harm are the ones nobody thought of, which is why they are missing from the code.
Another is asking the same AI tool that wrote the code whether it is secure. It will answer confidently, with the same blind spots it had while writing it.
A third is reviewing once and then carrying on prompting changes into the live app. Every change after a review is unreviewed until someone looks at it.
And the reverse mistake happens too: treating every vibe-coded tool as dangerous and paying for reviews of a rota or a dashboard only you use, when nothing in it could hurt a customer.
How We Approach This
We use AI in our own work and build it into systems for clients, so our view is a practical one. AI tools get you something approximately right, quickly. For a lot of jobs that is enough. Where it isn’t, the gap is closed by a developer reading the code against a written description of what it should do.
That is what our AI Sign-Off service does. A developer reviews the app against a short brief, fixes what fails and signs off the version they reviewed, against a stated scope. If you want us to stay accountable for the code as it changes, that continues on a support retainer.
Before It Goes in Front of Customers
If your app is about to take bookings or payments from the public, the time to get a developer to look at it is before launch, while it is still cheap to change and before anyone outside the business relies on it. Run the owner’s checks first, then decide whether you need a review. If you are already further along and know things need fixing, our guide on what it costs to fix a vibe-coded app explains what moves the price. More answers on related questions 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.