Artificial intelligence is increasingly being used to carry out cyberattacks, making them faster, more precise and harder to detect, Lithuania's National Cyber Security Centre has warned.
The warning comes in the centre's 2025 Cyber Threats Report, compiled with international partners, which found that AI is enabling both criminal groups and state-sponsored actors to carry out more sophisticated and large-scale attacks.
The technology allows attackers to gather information on targets more efficiently, craft convincing phishing emails and exploit security vulnerabilities more rapidly.
The report also highlights a shift in ransomware tactics, with hackers increasingly targeting cloud services and software-as-a-service (SaaS) platforms, where critical data is stored and day-to-day operations take place, rather than focusing solely on internal networks.
Of 5,418 recorded cyber incidents in 2025, two types of attack accounted for more than 84% of cases. Distributed denial-of-service (DDoS) attacks, which overwhelm systems with traffic to knock them offline, made up 44% of incidents, while ransomware, in which attackers encrypt a victim's data and demand payment for its release, accounted for 40%.
A significant share of DDoS attacks was carried out by politically motivated hacktivist groups, primarily targeting countries that support Ukraine. The centre said such attacks rarely cause lasting technical damage but are designed to attract attention, amplify hostile narratives and apply political pressure.
Phishing, in which users are tricked into revealing sensitive information through fraudulent messages, remains the most common method used to gain initial access to systems. AI is making such attempts increasingly difficult to identify, the report found, as even professionally worded messages may now be machine-generated.
The centre urged organisations to update their systems regularly, introduce multi-factor authentication and improve both cloud security and staff awareness. Members of the public were also advised to remain vigilant, as even professional-looking messages can be AI-generated.
Newsletter
Every Friday morning.
