In late July 2025, the women-only dating-advice app Tea went viral for all the wrong reasons.
Marketed as a safe, private space to anonymously flag men for “red-flag” behavior, hackers attacked the app, accessing over 70,000 user-submitted images including passport scans and driving licenses, and more than one million private messages.
The flaw was almost comical in its simplicity, as Tea’s backend was a public Firebase database with no encryption or authentication.
While many sympathised with the women affected, the conversation was soon hijacked by ridicule, much of it aimed at CEO Sean Cook for his lack of development experience. Tea became the prime example of the dangers of vibe coding, building fast, with too much help from AI, but without rigorous testing (or even basic security).
In developer circles, the incident sparked a conversation about the benefits and limitations of vibe coding. Is it useful only for rapid, low-stakes or internal projects, but potentially dangerous for complex, high-risk systems? Or can it scale in a secure way?
When Vibe-Coders Fail
A few weeks later, in early August, I spotted a Reddit comment from a self-described “real engineer” who said they enjoy watching vibe-coders fail, and it got me thinking: do other professional developers share this attitude?
Now of course plenty of developers vibe-code, but for the purposes of this article we’re referring to non-professional developers who are building with vibe-coding tools.
The tech community is very inclusive, but is this attitude a form of gatekeeping if we openly mock people who are new to building software, rather than mentoring or guiding them into learning more?
Could this attitude dissuade vibe-coders who might actually be interested in learning to code from actually doing it, because they’re shunned by real developers?
I posted on Threads to find out how developers feel about it, asking “Do ‘real’ developers enjoy seeing vibe-coders fail? I’m seeing this kind of attitude quite a lot. Is it just me or does it feel quite gatekeepy?!”
The replies were… illuminating, and show that there’s a real issue here we’ll inevitably need to address.
Learn to Walk Before You Run
This user’s response reveals that for some, the issue with vibe-coders is that they overlook the fundamentals of development, trying to do too much too soon.
If we follow this idea forward, we can understand why professionals might not show much sympathy when vibe-coders come unstuck. “I told you so” certainly comes to mind.
Understanding the Risks
As the Tea debacle showed, vibe-coding carries real-world risks. When you’re working on personal or sensitive data, understanding how to keep it secure is essential, not a choice.
For one user, gate-keeping might actually be necessary, as without understanding how to build software vibe-coders are doomed to build insecure, poor quality software rife with technical debt, a point echoed by Steve Krouse in his article about how vibe code is legacy code.
Respecting the Craft
Another user suggested it’s not actually gatekeeping at all, to have these kinds of attitudes. Instead, they suggest that seeing vibe-coders fail is satisfying only because of their blasé attitude towards software engineering in general.
This reframes the conversation to say that it’s not about shutting people out, but that in some way it’s satisfying to see people who underestimate the craft of programming find out what it’s all about.
AI Evangelists
AI evangelists talk a big game about the transformative effect of AI, and with so many bold claims made by so many about it, it’s difficult to know what is hype and what is to be believed.
To some, the hype and hyperbole around AI makes up for a lack of substance, suggesting AI is in general not very well understood - even by those working with it, let alone those who do not:
“I watch with indifference and very little sympathy as people trust these systems that are wrong so often and even the companies who build them can’t control. It’s so easy to ‘trick’ these models… they can’t ‘follow instructions,’ they just generate tokens. They’re inherently unreliable. Agentic AI with LLMs is absurdity at best.”
The Future Could Be Vibe-Coded
Not all experienced developers are opposed. One self-described “real coder” openly embraced the approach:
Perhaps vibe-coding isn’t a threat, it’s an evolution. Perhaps developers will focus more on higher-level problem solving than on manually writing code - a possible reality that developers old and new just need to accept, whether it’s popular or not.
Takeaways
As we said at the beginning, plenty of developers vibe-code. The difference is they understand the risks of doing so, and approach it differently than those with no technical background.
With that in mind, it’s an opportunity for developers to guide vibe-coders towards learning more about their profession.
It is understandable that some developers find the hype and conversation around AI exhausting or irritating, but it’s important we don’t discourage a potential developer from being one.
Vibe coders, remember to walk before you run, learn the basics (at least), respect the craft of programming, be mindful of scope and know when you’re out of your depth.
Developers, be understanding and inclusive, mentor and guide beginners away from preventable mistakes, teach good practices.