A national cybersecurity agency has stated that there is a "high likelihood" that AI-powered malware will be used in attacks on critical infrastructure such as power plants, airports and hospitals by 2028. The assessment points in particular to AI potentially enabling a ransomware attack against critical national infrastructure. The definition of critical infrastructure here covers the energy, transport, defence, food, health and finance sectors.
What happened?
The core of the assessment is simple: rather than moving cyberattacks into an entirely new category, AI is making existing types of attack more accessible and faster. The agency expects known threats such as ransomware to be strengthened by AI on a 2028 horizon. In other words, there is no single new weapon going off like a bomb; rather, there is a lowering of the threshold for an attack.
Why does it matter? The capability gap between attacker and defender
The truly striking point is not some devastating new tool, but a shift in the asymmetry. Until now, the most serious attacks on critical infrastructure have been the work of state-backed advanced persistent threat (APT) groups. These groups had the resources to find zero-day vulnerabilities, write custom malware and patiently infiltrate over months. It was difficult for the ordinary attacker to reach this level of capability. This is precisely the balance that AI is changing.
AI is rapidly raising the capability of an unskilled or mid-level attacker. Steps such as reconnaissance, generating phishing copy, code analysis, vulnerability scanning and producing malware variants, which previously required expertise and time, are now being automated. The result is that the capability gap between attacker and defender is closing in the attacker's favour. The threat no longer comes solely from elite groups, but from attacks that are becoming commonplace and scaling up. It is becoming possible to launch more convincing and more tailored attacks against large numbers of targets at low cost.
This is why the date of 2028 is not arbitrary. Generative AI tools are maturing, their use on the attacker's side is spreading, and the defensive side has not yet fully kept pace. The three-year horizon appears to reflect a reasonable lag between today's capability curve and its translation into actual attacks.
Why is critical infrastructure the most fragile link?
What makes critical infrastructure especially vulnerable is its often outdated and slowly patched operational technology (OT) and SCADA systems. Updating a server on an office network is relatively easy; updating a controller at a power plant or water treatment facility often requires halting production. These systems are designed to run for decades, with continuity prioritised over security.
| Dimension | Enterprise IT | Critical infrastructure (OT/SCADA) |
|---|---|---|
| Patching speed | Days/weeks | Months, sometimes years |
| Priority | Data security | Continuity, human safety |
| System lifespan | 3-5 years | 15-30 years |
| Attack impact | Data loss | Physical disruption, public order |
The lag shown in this table leaves a wide window of opportunity in front of an attacker accelerated by AI. While the attacker produces a new variant within minutes, the defending side's patch cycle stretches over months. This difference in speed makes critical infrastructure the weakest link in the chain.
What's next? A race in which both sides use AI
The picture is not one-sided. The defensive side is also using AI for anomaly detection, automated incident response and threat hunting. In the period ahead, the key thing to watch is which way the balance will tip in this arms race. The attacker's advantage is speed and scale; the defender's advantage should be knowing its own network better than the attacker does.
Several themes to watch are coming into focus: whether segmentation and network separation will become widespread in OT systems, how much supply chain security will tighten, and whether regulation can keep up with this pace. In the case of ransomware specifically, the backup and recovery capacity of critical services will be decisive.
Ultimately, this warning is not a prophecy but a trend-based risk assessment. Rather than creating panic, it says this: what is changing is not the nature of the threat, but its accessibility and speed. For critical infrastructure operators, the real test will be waged not against a new weapon, but against their own slow patching culture. For this reason, the assessment should be read less as a technical warning and more as a call to re-order priorities.
Yorumlar (0)
Henüz yorum yok. İlk yorumu sen yaz.
Yorum yapmak için Sinyal'i indir
Yorumlar Sinyal hesabıyla yapılır. Mobil uygulamada giriş yap, yorum bırak.