AI platforms are transforming many industries. But perhaps none more so than software development. “Vibe coding” was only coined as a term in February 2025. Yet just a few months later, one report suggested 84% of developers were using or planning to use AI tools for work.

On paper, it’s obvious why they are doing so. AI does the heavy lifting, allowing the developer to let their creativity flourish. But in so doing, vibe coding tools also lower the barriers to entry for novices unable to spot bugs and mistakes. For some time-poor developers, the technology may also provide a false sense of security. These oversights aren’t necessarily going to be flagged by the platforms on which the software is distributed.

All of which puts the onus on users to vet their apps more carefully than perhaps they did in the past. But what are the risks to look out for, and what are the right questions to ask?

Common vibe coding mistakes

Vibe coding tools are designed to prioritize functionality, and look and feel, over quality. That can lead to some concerning oversights creeping in. These might include:

  • Hardcoded secrets such as API keys left in the app’s code, which malicious actors can extract, sometimes automatically. It could let them into the developer’s backend and the user data stored there, including yours.
  • No input validation or access controls, which could expose an app to accepting malicious input (data), or enabling users to access or change data belonging to other users.
  • Public-by-default settings which could allow users to view the profiles or private information of other users of an app.
  • Weak or missing encryption, which makes data stolen from an app, or intercepted on its way to and from it, easier to read and exploit.
  • No rate limiting, which means hackers could carry out “brute-force” attacks, using automated software to guess your password a huge number of times.

Vibe coded apps may also be exposed to prompt injection. This is a type of attack targeting AI tools where hackers hide specific malicious instructions in their prompts, or in content on the web that the tool processes. If the AI has access to your private data and accounts (e.g., a personal assistant like OpenClaw) it could spell trouble.

A cautionary tale

Unfortunately, these potential mistakes aren’t theoretical. App users have been exposed in the past to security and privacy risks because of coding oversights their developers made.

A good example is vibe coding platform Lovable. One security researcher found 16 vulnerabilities in a single app hosted on the platform, six of which were reportedly rated critical. Some exposed sensitive user information. The app in question had garnered more than 100,000 views. Lovable said it has since fixed the issues.

What can go wrong with vibe coded apps?

The above list is not exhaustive. But more important than the technical details is what coding mistakes can lead to if you download the wrong app. It amounts to data loss, financial exposure and potential malware installation. A poorly coded app might:

  • Store your passwords or session tokens in an insecure way, exposing you to account or device compromise
  • Mishandle your personally identifiable information (e.g., email and home address, and ID details) enabling hackers to perform identity fraud
  • Allow other users to view your data, which could be a fraud or privacy risk
  • Expose your AI prompts and any sensitive data within those
  • Leak your payment information, putting you at risk of someone draining your account

What are the right security questions to ask about vibe coded apps?

The challenge is vetting the apps that you come across. Some marketplaces are more rigorous than others in terms of the checks they apply to software distributed through their platforms. So just because it’s listed, it doesn’t mean it’s safe. But similarly, just because it’s vibe coded, it doesn’t mean it’s dangerous.

With that in mind, here are a few questions that might steer you in the right direction:

  1. Who is the developer? Is it a legitimate company? How long have they been around and what are their reviews and reputation like? Dig into any negative reviews and see what they say about security or privacy. If it’s a viral sensation that’s blown up overnight, it may be a risky bet. Look for a legitimate-looking support channel to help you in case something goes wrong.
  2. What is it asking for? Check what permissions and data access the app requires. Do they seem appropriate for the type of app? A calculator that requires access to your camera would be a no-no, for example. Remember: the more sensitive data it can access, the bigger the potential risk.
  3. What does the privacy policy say? If there’s one at all, does it clearly explain what data it collects, where it’s stored, who it shares that data with, and how long it’s retained? Similarly, what are its data deletion policies if you decide to quit the app?
  4. What is the security model like? Apps that explain how they protect passwords and sensitive data, and have a mechanism for vulnerability reporting and security updates, are immediately more trustworthy, although this is just a baseline.
  5. If it uses AI, what can it access? This is important to understand how exposed you may be to prompt injection. The more permissions the AI has – to access sensitive data, and perform actions like sending messages and making purchases – the greater the risks if it is hijacked.

What to do if I already used a breached app?

If it’s already too late, there are still some steps you can take to contain the threat.

With a badly coded app, the leak usually happens on the developer’s servers, so start with your account and credentials. Change the app’s password if you can still log in, then request account deletion – uninstalling the app doesn’t delete the data it holds on you. Change your logins across any other apps and sites that share the same password, and turn on multi-factor authentication. Revoke any connected account permissions, such as Sign in with Google or Apple. And monitor your bank accounts and other online accounts for identity misuse, such as unfamiliar orders made in your name.

If you suspect the app itself is malicious, uninstall it. On Android, run a scan using trusted security software. If you can’t remove the app, consider performing a factory reset, and make your password changes from a different device.

Frequently asked questions (FAQs)

What does 'vibe coded' mean?

That the app was built largely by describing it to an AI tool in plain language and accepting the code it generates, often with little or no review. It lets almost anyone become a developer.

How do I know if an app was built with AI?

You usually can’t tell and there’s no easy way to find out. It’s better to check for other things such as security model, privacy policy and developer reputation.

Are all vibe coded apps dangerous?

No, it depends on the developer and the platform. However, if it’s a novice coder they may have made some rookie errors that put users at risk.

How do I know if an app is dangerous or not?

Find out information such as who the developer is, what their reviews are like, what permissions the app asks for, what the security model and privacy policy say, and what it’s AI capabilities can access.

What if I’ve already used a compromised app?

Change the app’s password, request account deletion and uninstall it. Change your password on all other apps and sites where you use the same credentials, and turn on multi-factor authentication. Revoke permissions for connected apps. Monitor your bank and other accounts for misuse. If you suspect the app is malicious, scan your Android device and consider a factory reset if that doesn’t work