Opinion: AI-assisted attacks demand runtime behavioural detection and enforcement for applications and APIs

The author argues that AI-assisted attackers can probe, adapt and chain actions mid-attack, which undermines defences that rely on signatures, WAF rules, API schemas and threat intelligence. The piece recommends runtime behavioural detection across live application and API traffic, linked to graduated enforcement such as blocking, rate limiting or tightening policy, applied consistently across hybrid multicloud, Kubernetes and edge environments. It cites Anthropic◼'s 2026 analysis of 832 banned accounts and research finding 48% of organisations update production APIs daily or more often while only 19% keep a fully automated, continuously updated API inventory.

For IT people · Written by AI from 1 source · Wrong?