Multi-turn attacks broke AI models 88% of the time — single-turn testing missed it, Cisco AI security lead warns at VB Transform 2026

In a recent study, Cisco discovered that multi-turn attacks can break through AI models 88.3% of the time, a stark contrast to single-turn testing that failed to detect such vulnerabilities. Amy Chang from Cisco highlighted this alarming statistic at VB Transform 2026, emphasizing that organizations relying solely on single-turn red-teaming are at significant risk. This finding underscores the urgent need for more adaptive security measures to protect against evolving AI threats.