Does AI make cyberattacks worse?
Criminal groups and state-backed hackers already use AI, mainly for phishing, reconnaissance and finding software flaws. Whether this has raised the overall level of attacks is still unclear, and AI is also used for defence.
What has been observed
According to the International AI Safety Report 2026, there is now strong evidence that criminal groups and state-sponsored attackers use AI in their operations. Whether AI has increased the overall scale and severity of cyberattacks remains uncertain, because cause and effect are hard to establish.
AI is particularly good at finding software vulnerabilities and writing malicious code. In one major competition, an AI agent found 77% of the vulnerabilities in real software and placed in the top 5% of more than 400 mostly human teams. At least one real attack has used semi-autonomous AI with humans stepping in only at key decisions, but no fully autonomous attack from start to finish has been reported.
What security agencies expect
The UK's National Cyber Security Centre (NCSC) judged in 2024 that AI will almost certainly increase the volume and impact of cyberattacks, mainly by making phishing and reconnaissance more effective and by lowering the bar for less skilled criminals. In its 2025 assessment looking ahead to 2027, it named AI-assisted discovery of vulnerabilities as the most significant development. The time between a flaw becoming known and attackers exploiting it has already shrunk to days, and AI will likely shorten it further.
The NCSC expects a growing gap between organisations that keep up and those that do not, and sees a realistic possibility that critical systems, especially older industrial control systems, become more exposed to advanced attackers by 2027.
Defence uses AI too
The same capabilities that help attackers find flaws also help defenders find and fix them first. The International AI Safety Report describes this as a dual-use dilemma: blocking AI from answering security questions altogether would also hamper defenders. AI companies now regularly report attempts to misuse their systems in cyberattacks and shut down the accounts involved.
What is being done
Security agencies stress the basics: patch quickly, use multi-factor authentication, keep offline backups and segment networks. Under the EU AI Act, high-risk AI systems must meet requirements on robustness, and providers of the most capable general-purpose models must assess and mitigate systemic risks.
What you can do
- Install updates on your phone, computer and router promptly; many attacks exploit flaws that already have a fix.
- Turn on two-factor authentication for email, banking and social media.
- Assume that a well-written, personal message can still be phishing; check the sender through another channel before clicking.
- If you run an organisation: keep an inventory of your systems, patch fast and keep backups that are disconnected from the network.
Sources
- Impact of AI on cyber threat from now to 2027 – National Cyber Security Centre (UK), 2025
- The near-term impact of AI on the cyber threat – National Cyber Security Centre (UK), 2024
- International AI Safety Report 2026 – International AI Safety Report (chair: Yoshua Bengio), 2026
- Inga större informations- eller cyberangrepp mot svenska valet – Myndigheten för psykologiskt försvar (MPF), 2026
- AI Act – regulatory framework for AI – European Commission, 2026
Data updated: 2026-11-19