AI accelerates cyberattacks, threatening daily life

Summary

El País reports that AI has dramatically shortened the time between the discovery of security vulnerabilities and their exploitation. This period has dropped from 63 days in 2018 to approximately one week in 2025. Recent attacks on supermarket refrigerators in 12 European countries and water systems in the US demonstrate that cyber threats are now causing concrete, physical consequences. The article frames this situation, where the boundary between the digital and physical worlds is eroding, as a 'paradigm shift.'

Why it matters

The acceleration of attack cycles makes it impossible for defenders to rely on the historical patch window for protecting critical infrastructure. This fundamentally changes the economics of cybercrime. As the article notes, a corrupted refrigerator cannot be 'recovered' like a database, shifting risks from financial loss to food safety and public health. The democratization of attacker tools via AI enables smaller actors to target systems previously reserved for nation-state rivals.

Highlights

• According to Andreessen Horowitz data, the exploitation time dropped from 63 days in 2018 to approximately 1 week in 2025
• Researchers from King's College and Kaspersky state that AI scales zero-day discovery, exploit code generation, and target research
• Cyber-physical convergence means attacks now lead to irreversible physical damage (corruption, equipment destruction) rather than just data loss

Critical take

The article's causal framework is somewhat reductionist: it largely attributes the current wave of attacks to AI. However, the examples cited—Stuxnet, Industroyer, and the 2014 Jeep hack—occurred a decade before generative AI. The fundamental security vulnerabilities in cyber-physical systems are architectural in nature and not specific to AI. Furthermore, the article bases its 'new era' thesis on a small group of experts, failing to address the counter-argument that the observed acceleration stems from advanced tools and market dynamics rather than a qualitative AI breakthrough.