OpenAI creates an artificial intelligence specialised in cybersecurity: this is how it protects you
OpenAI, the same company behind ChatGPT, has announced something that may sound strange but is actually good news for you. It has created an artificial intelligence program (a kind of digital “brain” capable of learning and reasoning) specially trained to look for security flaws in computers and programs. It is called GPT-5.6-Cyber and it is designed to help the “good guys”, that is, the technicians who work to protect people and companies from cybercriminals.
And why does this matter to you? Because the more flaws are found and fixed early, the less likely it is that a criminal will sneak into your data. It is like checking the locks of your house every day to fix any weak door before a thief finds it. OpenAI released this new feature on 10 August 2026.
What exactly is this new tool?
To understand it, think of ChatGPT as a general assistant that knows a bit about everything. GPT-5.6-Cyber is like a specialist: it is not a “jack of all trades” assistant, but an expert with a single mission: finding weak points in software (that is, in the programs we use every day, such as the internet browser, mobile apps or the databases where data is stored).
The creators themselves tell a surprising fact. In internal tests, this specialised brain responds well to 95% of the advanced security tasks asked of it. The general program, the one most people use, barely reaches 1.5%. It is a huge difference, like comparing a professional firefighter with someone who has only seen a candle burn.
It has already found real flaws, including in your browser
This is not theory. OpenAI assures that this artificial intelligence has already found real flaws in very widely used programs. For example, it has detected two hidden problems in the engine that runs Chrome, Google’s browser that millions of people use every day. Google’s people have already fixed one of them.
It has also helped locate at least five flaws in a very popular mobile operating system (the program that makes a phone work), three serious problems in a well-known database (a place where enormous amounts of information are stored) and more than 400 flaws in the “heart” of an operating system. Those are a great many weak doors that have been closed before anyone malicious could take advantage of them.
A security flaw is like an open window in your digital house: someone could get in without permission. Finding those windows before the thieves do is exactly the job of these tools.
Can anyone use it?
Here comes the important and reassuring part: no. This very powerful tool is not for sale to everyone nor is it available in the ChatGPT you use. Only security companies and technicians who pass a very strict vetting can access it, proving that their work is legal and that they want to use it to defend, not to attack.
It is a logical decision by OpenAI. A tool like this, like a very sharp knife, can be used to cook or to cause harm. That is why it is only given to trusted people and with special permission, designed to protect important infrastructure or to monitor their own clients’ programs.
What does this mean for you?
For the average person (like you or me), this announcement translates into something simple: companies are using artificial intelligence to make your programs, your browser and your applications more secure. Every flaw found and fixed early is one less risk to your passwords, your photos and your personal data.
What you can do is make a habit of keeping your devices updated. An update is like a vaccine for the computer or phone: it fixes weaknesses that criminals could exploit. When you get the notice that there is a new update, do not put it off; let it install. It is one of the easiest and most effective ways to protect yourself.
In short: artificial intelligence not only creates things or chats with you. It can also side with the good guys and help make your digital world a little safer, even if you do not notice. And that, in these times, is excellent news.
