The rise of AI-powered cyberattacks: new risks and challenges for cybersecurity

  • AI powers attacks like phishing, ransomware, and sophisticated malware.
  • Modified language models are used to create threats that are difficult to detect.
  • Organizations must adapt their security measures by monitoring traffic and training their staff.
  • Automation and social engineering make cyberattacks increasingly effective.

Cyberattacks and artificial intelligence

In today's digital age, artificial intelligence is transforming how cyberattacks are carried out . What was once the domain of computer experts is now within reach of criminal groups thanks to AI systems capable of developing and refining threats with unprecedented speed and effectiveness. This new reality presents a context of increasing risk for companies, public organizations, and individual users, who must urgently adapt to this challenge.

Far from acting solely as a defender , AI has become a double-edged sword in the world of cybersecurity. Advanced language models, once associated with productivity and innovation, are now being used to automate attacks, build customized phishing campaigns, and design increasingly difficult-to-detect malware .

The role of AI models in digital crime

Cybersecurity and AI attacks

According to recent analyses by cybersecurity experts, cybercriminals are leveraging both publicly available and modified versions of AI to circumvent ethical and legal restrictions. These tools enable the creation of flawless fraudulent messages or malicious code without errors, increasing the success rate of attacks and drastically reducing the margin of error.

Some models, such as Ollama and WhiteRabbitNeo, have garnered attention for their offensive and defensive capabilities, and for being adapted by criminal groups to bypass security filters. Furthermore, customized versions hosted on the dark web —such as GhostGPT, WormGPT, and FraudGPT—allow for the generation of everything from Trojans and ransomware to the automated verification of stolen credit cards or the detection of vulnerabilities in websites.

This rapid development is largely due to the ability to modify training data and model parameters , eliminating barriers that prevent dangerous or illegal responses. Not only are attack methods being reinvented, but criminals are also sharing and selling these tools on underground forums, facilitating their proliferation.

Sophistication and automation in cyberattacks

Artificial intelligence has not only accelerated the development of threats, but has also enabled the application of automation and social engineering techniques at scale. Mass phishing emails, identity theft, and data manipulation are now carried out with pinpoint precision and alarming results.

Innovative attacks based on manipulating the models themselves have also emerged , for example, through prompt injection to bypass the protection filters of legitimate systems. The combination of AI and malware has led to increasingly intelligent ransomware , capable of encrypting data, evading defenses, and self-destructing after completing the attack, as evidenced by recent operations attributed to groups like FunkSec.

Cybercriminals employ low-cost strategies to maximize the reach and profitability of their campaigns, ranging from offering relatively low ransoms to the direct sale of stolen data on underground markets, which fuels new waves of attacks.

Strategies to address the AI-driven threat

Responding to this wave of attacks requires adapting to new threats. Experts recommend strengthening defenses and monitoring AI-related traffic in real time, identifying unusual activity, and, above all, training staff to detect signs of attacks driven by language models.

Among the most effective strategies are:

  • Keep systems and applications up to date, as attackers often exploit unpatched vulnerabilities.
  • Implement offline backups and anti-ransomware systems that make it difficult to recover files in the event of an attack.
  • Monitor access and information exchange involving AI tools, looking for abnormal patterns.
  • Train employees to identify phishing emails generated by language models.
  • Use advanced detection solutions that are capable of recognizing techniques used by AI.

Collaboration between the public and private sectors—as well as the use of trusted cybersecurity platforms—is becoming essential to intensify the fight against AI cyberattacks and curb the proliferation of these threats.

The future of AI-powered cyber defenses

The battle is no longer just between hackers and human defenders, but also between increasingly complex algorithms . Experts predict that the only way to meet these challenges will be by incorporating AI into defenses themselves, using models that detect anomalies, monitor traffic in real time, and help contain the spread of malicious activity.

This "AI versus AI" dynamic will lead to a scenario where constant vigilance, updated strategies, and ongoing training will make the difference between falling victim to a devastating cyberattack or maintaining resilience in the digital age. Furthermore, it will be key to anticipating and neutralizing these emerging threats.

The presence of modified and customized AI models in cybercrime compels both businesses and individuals to adopt a more proactive and informed approach. As threats evolve at the speed of technology, only a combination of technical measures, training, and constant vigilance will allow us to manage the growing risks. Artificial intelligence is here to stay; now we must decide how to coexist with its darker side and take action to avoid falling behind.

electronic warfare Ukraine drones
Related article:
Latest developments in electronic warfare with drones in Ukraine

Add as preferred source in Google