Artificial intelligence is rapidly transforming the technology industry. Today, a simple prompt in ChatGPT, Claude, or Gemini is enough to create a website, an online store, or even a complete management system.
But what many users fail to consider is that the same artificial intelligence used to create a website can also be used to discover its vulnerabilities and hack it.
And a recent incident involving Google shows that this risk is no longer just a theoretical scenario.
Gemini Hacked 3 Real Companies During a Security Test
On September 18, 2026, Google confirmed that its artificial intelligence model, Gemini, had gained access to the systems of three real companies during a cybersecurity test conducted in May.
According to The Wall Street Journal, the test was being carried out by Irregular, a company specializing in evaluating the security of AI systems.
Gemini had been tasked with testing the security of a simulated company. However, due to unintended internet access, it ended up interacting with real-world systems.
In one case, the AI managed to discover the password of a protected system. In the other two cases, it found credentials that had been publicly exposed online and used them to gain access to the companies' systems.
Google stated that Gemini stopped its actions after identifying that the companies were not part of the simulation. The affected companies were notified of the incident.
Are We Creating Websites Today That AI Could Hack Tomorrow?
This incident raises a question that every business should be asking itself:
If artificial intelligence can build a website in just a few minutes, how long would it take another AI to discover its vulnerabilities?
AI models have been trained on vast amounts of programming code and can recognize common structures, patterns, and mistakes in application development.
When a website is created entirely using AI, without professional oversight, its code may contain security vulnerabilities that are not visible to the average user.
Another AI system can be used to analyze the application, discover weaknesses, and identify ways in which those vulnerabilities could be exploited.
A Beautiful Website Doesn't Mean a Secure Website!
One of the biggest problems with AI-powered website development is excessive trust in automatically generated code.
A website may feature a modern design, animations, an administration panel, and advanced functionality, while serious security vulnerabilities remain hidden within its underlying structure.
If the code is published without proper review, databases, administrator accounts, and customers' personal information could be left exposed.
Furthermore, the risk is not limited to websites alone.
AI is increasingly being used to create applications, booking platforms, payment systems, and software that manages sensitive information.
A small programming mistake can have serious consequences when a system is used by thousands of people.
The Technology Paradox: AI Writes the Code, AI Finds the Vulnerabilities!
In the past, developing and testing an application required considerable programming knowledge.
Today, artificial intelligence is making both processes significantly easier.
Someone with no programming experience can use AI to create a complex website. Similarly, an attacker can use the same technology to search for vulnerabilities in publicly accessible systems.
This creates an unusual situation: the technology making programming accessible to everyone is also making security analysis tools more accessible.
The incident demonstrates that advanced AI models can already identify and exploit real-world vulnerabilities without requiring detailed instructions at every step.
Should We Trust AI to Build Our Websites?
Artificial intelligence is a powerful tool for developers and businesses, but it should not be considered a complete replacement for professional expertise.
Building a website does not end the moment it goes live and starts functioning.
Database security, personal data protection, server configuration, access control, and software updates are essential elements that require continuous attention.
Just because anyone can build a website with AI today doesn't mean anyone can guarantee its security.
The Gemini incident is a clear reminder that artificial intelligence is changing not only how the internet is built, but also how it can be attacked.