UK must expect AI-powered attack on critical infrastructure within two years, report says | Reuters

Open original

The UK's National Cyber Security Centre says Britain should prepare for an AI-powered cyberattack on its critical national infrastructure within the next two years. It also judges China to be the state best equipped to carry out such operations.

  • The NCSC expects an AI-enabled attack on critical national infrastructure within roughly two years, a near-term timeline rather than a distant risk.
  • China is named as the actor most likely to be able to mount these operations.
  • The warning is framed as a call to prepare now, with the government treating AI-driven attacks as a planning assumption.

The core claim is straightforward: the UK's National Cyber Security Centre (NCSC) told the country to prepare for an AI-powered cyberattack on its critical national infrastructure within the next two years. The statement came on a Tuesday, and the article is dated November 3, 2026. The agency also named China as the state most likely to have the capability to carry out such operations, which is the detail that will draw the most attention from policymakers and the press. The reporting is thin on the specifics that would let a reader judge the threat for themselves. The excerpt I have doesn't say which sectors the NCSC considers most exposed, such as power grids, water, telecoms, or transport, nor does it describe what an "AI-powered" attack would look like in practice. Is the concern autonomous malware that adapts to defenses, AI tools that speed up reconnaissance and exploit development, or something else? The short piece leaves that open, so if you want to understand the risk rather than just note the headline, the NCSC's original publication is the place to look. The mention of China is also a flat attribution in a single sentence, with no evidence cited in what's available here, so treat it as the agency's assessment rather than established fact. Still, the timeframe matters. A two-year warning is a fairly concrete planning horizon, and it signals that the NCSC sees AI as moving from a general productivity tool to something that could meaningfully change how state-level intrusions get run. For anyone working in critical infrastructure operations or security, that points toward reviewing exposed systems and patch cycles now rather than waiting for the threat to mature. For everyone else, the takeaway is that a national security agency has put a date on it.

Ask about this article