DIGITAL LIFE

AI is already changing the way cyberattacks unfold

A new report reveals that criminals are already exploiting artificial intelligence models in unprecedented ways. This scenario concerns experts and could completely transform how companies protect their systems.

For a long time, artificial intelligence was viewed primarily as a tool to boost productivity, create content, and automate tasks. However, as these technologies evolve, they are increasingly being adopted by criminal groups for ever more sophisticated attacks. A recent report shows that this shift is already underway, indicating that cybersecurity may be entering a new era where intelligent programs face off against other equally intelligent programs.

A study highlights that language models are being used by cybercriminals in far more advanced ways than imagined just a few years ago. Rather than merely accelerating repetitive tasks, artificial intelligence has begun to play a direct role in decision-making during cyberattacks.

In practice, this means certain malicious programs can alter their own structure while running, modifying code segments to evade detection by traditional security tools.

This behavior marks a significant shift from conventional malware. Historically, antivirus software and protection platforms relied on analyzing digital signatures, known patterns, and previously cataloged behaviors to identify threats. Now, however, some programs can continuously alter these patterns, making detection far more complex.

According to experts, some groups also use language models to automatically create new routines, adapt strategies based on the target system's response, and modify execution methods to bypass defense mechanisms.

While this type of technology does not yet fully replace human involvement, it reduces the time required to plan attacks and enhances adaptability during an intrusion.

Another aspect highlighted in the report is the growing use of AI by cybercrime groups and state-sponsored organizations, demonstrating that the technology has moved beyond the experimental stage to become an integral part of actual operations. The advancement of artificial intelligence has also brought about another problem: criminals are learning to exploit AI models themselves to obtain information that facilitates the creation of malicious code.

Major commercial systems typically have security mechanisms in place to reject dangerous requests. Even so, researchers observe that some users attempt to bypass these barriers by presenting seemingly legitimate requests, such as academic exercises or information security research. This type of manipulation—known as AI-based social engineering—has become a new area of ​​focus for the companies responsible for these models.

Furthermore, experts warn of a growing underground market for AI-based tools sold to facilitate attacks such as phishing campaigns, the automated production of malicious code, and other illegal activities.

This scenario reinforces a trend already under discussion within the industry: the future of cybersecurity will increasingly depend on the use of artificial intelligence for defense as well.

Instead of relying solely on systems based on fixed rules, companies are investing in platforms capable of continuous learning, identifying unusual behaviors, and responding rapidly to novel threats.

The expectation is that the next generation of security solutions will utilize intelligent models to analyze vast amounts of data in real time, detect suspicious patterns, and react before an attack causes significant damage.

Ultimately, the transformation lies not only in the evolution of malware but in a shift in the digital battlefield itself. While the conflict once played out between hackers and traditional antivirus software, all signs point to a future confrontation between AIs trained to attack and those developed to thwart such attacks.

This technological race is just beginning, and its evolution could determine how companies, governments, and users protect their data in the years to come.

AI has already changed the nature of attacks by democratizing cybercrime, reducing execution time from days to seconds, and enabling large-scale intelligent automation. Advances in frontier AI and autonomous AI have redefined the digital threat landscape.

1. Lowering the barrier to entry...Attacks that once required advanced programming knowledge can now be executed by inexperienced criminals. Generative AI models can be instructed via simple prompts to write malware code, debug scripts, or structure entire campaigns.

2. Machine-speed attacks...AI analyzes thousands of lines of code, networks, and APIs for vulnerabilities at lightning speed, far surpassing human checking capabilities. Where humans once spent hours or days finding a flaw, AI takes only minutes.

3. Hyper-realistic social engineering...The classic phishing email riddled with grammatical errors is a thing of the past. AI now generates perfectly personalized and convincing text. Through voice cloning and video deepfakes, criminals can highly convincingly impersonate executives or technical support staff.

4. Polymorphic and autonomous malware...Criminals employ AI agents that operate continuously. They create malware capable of constantly altering its own code (polymorphism) to evade traditional defense systems. Furthermore, AI can navigate corporate networks, validate data, and exfiltrate it without human intervention. To combat a landscape where AI attacks and defends at the same speed, companies and institutions must adopt automated defenses and behavior-based intelligence, while also protecting their AI models against manipulation. You can follow key cybersecurity trends in the AI ​​era via Palo Alto Networks or read more about the impact on organizations from Microsoft. The continuous operation of AI-powered bots, which work tirelessly, represents another advantage for attackers. Threat actors use generative AI to craft phishing lures, translate content, summarize stolen data, and generate or debug code. AI tools are already capable of independently identifying thousands of critical vulnerabilities in operating systems and browsers.

mundophone