Human Error Remains at the Core of AI-Enabled Social Engineering
KnowBe4, Monday, August 24th, 2026
KnowBe4 relays Resilience data showing AI-enabled social engineering drove 85% of cyber insurance losses in H1 2026.
AI is making social engineering attacks significantly more effective, according to a new report from cyber insurance firm Resilience. Such attacks were behind more than 85% of losses in the first half of 2026, compared with less than 20% during the first half of 2024.
That shift over two years is dramatic, and it is measured in paid claims rather than survey responses, which makes it a hard signal about what is actually costing organizations money.
The post examines what changed to produce the increase and why human error remains the mechanism even as the tooling around it improves.