AI-Enabled Breaches Rose 56% Year-Over-Year, New Data Shows

One in four data breaches between March 2025 and February 2026 was AI-enabled, up 56% from the year prior — a real, quantified acceleration, not a directional impression. In the first half of 2026 alone, there were 1,803 reported data compromises, compared with 1,732 in the same period the year before.

Why "AI-Enabled" Is a Meaningful Distinct Category

This isn't counting attacks against AI systems (like the incidents covered elsewhere in this batch) — it's attacks where AI capability itself made the exploitation easier or faster for the attacker, spanning AI-generated phishing content, AI-assisted vulnerability discovery, and AI-written exploit scripts (as seen directly in the Siemens PLC advisory covered separately this same week). The 56% year-over-year growth in this specific category suggests attacker adoption of AI tooling is accelerating faster than defensive tooling is keeping pace, at least by this measure.

• This data point provides real, measured context for individual incidents covered elsewhere this week (the Siemens PLC advisory, the Rust supply-chain attack) — they're not isolated anomalies, they're part of a measured, accelerating trend
• Security budgets and staffing plans built on pre-2025 threat assumptions are increasingly working from stale baselines given this rate of change — worth an explicit reassessment against current data, not historical planning assumptions