If you ask any cybersecurity expert (starting with the National Police itself), they will surely agree on one of the biggest threats that currently exist. It is none other than phishing. That is to say, scams that use identity theft to deceive users. The problem? That now, due to AI, its risks seem much greater.
That is at least the conclusion reached by none other than Microsoft itself. In a recent report, the technology giant once led by Bill Gates has tried to show how artificial intelligence is a boon for cybercriminals. And he has done it not with conjectures and opinions, but in the most forceful way that exists: with data.
The dangers of AI applied to phishing, according to Microsoft
It is not the first time that someone related to technology has warned about the same thing: artificial intelligence has given wings to cybercriminals. In fact, not too long ago a chilling figure emerged: the most common scams, the ones that everyone receives on their phone, had multiplied by 148%. Something that Microsoft seems to very much agree on.
The report in question, which responds to the title of Digital Defense Report 2025, indicates that many cybercriminals They already use AI to enter a network, move within it and end up launching an attack. Something that they are already capable of doing in just 12 days. Does it seem like a lot? Well, just one detail: before the arrival of AI, it took an average of around 58 days for the same thing.
However, the investigation decides to focus on what seems to worry experts the most: phishing cases. Apparently, when these are generated by AI they are much more dangerous. Its success rate is no longer around 12% (as happens in traditional cases) and shoots up to 54%. That is, more than half achieve their goal: they will deceive people.
The most complicated thing about the matter, they say, is because almost no one is prepared for these threats carried out with the help of artificial intelligence. From Microsoft they explain that antimalware, firewalls or patches are no longer enough; That is, more or less traditional measures, which seem ineffective to confront AI. If this is true, what can be done then?
Fight AI with AI
The only way to achieve a secure position against all phishing attacks created with AI, they say from Microsoft, is to also bet on this technology. Or put another way: resort to so-called defensive AI. That is, to systems that use AI for detection, automatic response, monitoring of anomalous patterns, rapid revocation of suspicious access, etc.
Still, the truth is that cybercriminals always give the impression of being one step ahead. And you don’t need to go to Microsoft to realize that this is the case. In Spain, people like INCIBE or even the DGT, who every now and then are suffering from this type of criminal strategies, have said exactly the same thing. The disadvantage is obvious.