Today, More than half of the spam – The 51% to be exact – It is generated by artificial intelligence(TO THE). To say it is a joint research of Columbia University, the University of Chicago and the Cybersecurity Barracuda company, which has Analyized millions of e-mails between 2022 and 2025. Compared to a year ago, when the spam generated by the AI was 40%, and to 2023, where it was even under the psychological threshold of 10%, the growth was exponential. Behind these increasingly refined and well-written messages, sometimes annoying advertising messages are hidden and, more and more often, attempts to very dangerous IT attacks, such as phishing or compromise of company emails.
Because the AI facilitates spam
The reason why artificial intelligence has become thus present in the spam phenomenon is not difficult to guess. One of the strengths of the AI is the ability of Write correct texts from a grammatical point of view, formal and convincingsometimes indistinguishable from those written by a person. The IT criminals understood it and, rather than reinventing their strategies, have decided to use advanced linguistic models-those at the base of the chatbots and virtual assistants-to finish their e-mail messages. The result? More “clean” spam, credible and difficult to recognize as such.
This new wave of intelligent spam represents an important evolution compared to the past. At one time, many unwanted messages were easily identifiable due to beating errors, uncertain grammar or suspicious tones. Now, however, these obvious signals are in fact disappearing. The e-mails created by the AI are able to adapt better to different linguistic culturesOf play professional and to transmit that urgency which pushes many users to click too quickly. The researchers explain that, like a human scammer, even an artificial intelligence knows well that urgency is a powerful psychological lever: those who receive a message that imposes an immediate reply is more inclined not to reason and act impulsively.
In the case of corporate e-mails, the phenomenon is particularly insidious. The so -called Bec (Business e-mail compromise), or compromises of the company mailboxes, exploit these techniques to pretend to be colleagues, customers or suppliers. According to the study, 14% of these attacks in 2025 were enhanced by the use of artificial intelligence. Although it is not yet the majority, the growth is constant. And since the access barrier to these tools is increasingly lower, an additional surge should be expected.
The other key element is the speed. A scammer can Create dozens of e-mails per daybut an artificial intelligence model can generate hundreds in a few minutes, all different, all to measure to overcome the antispam filters. The same tools used in marketing to improve newsletter texts are now used to refine the scams: more suitable synonyms, more effective tones, linguistic formulations designed for the target. The border line between tools for efficiency and abuse tools has become extremely thin.
Wei Haolearning at the Columbia Universityis one of the researchers who have worked on the study and, regarding the use of AI to generate e-mail spam truffaldine, explained:
Urgency is a tactic deliberated commonly used to exercise pressure and arouse an answer without thinking about the recipient. The attackers mainly use the IA to refine their e-mails and possibly their English rather than changing the tactic of their attacks.
How to defend yourself from the spam generated by the AI
As you can defend yourself From all this? Technological solutions, such as i Antispam filters oi Antivirus softwarehowever they are useful tools (not to mention essential), they are not enough, since you cannot prevent all Spam e-mails to reach the inbox of your mailbox. For this you have to invest as much as possible in what must be your first “defense barrier”: your computer skills. Learn recognize the signals of a possible attempt to scam – as urgent requests or the invitation to share personal data, banking information, and the like – is in fact the basis from which you have to start. Also because the future of the spam will be increasingly automated and-scratch messages will be increasingly insidious and credible, this is little but safe.
