Live test shows router rules fail against prompt injection; guard models needed

In his AustinLangChain talk, colinmcnamara live-tested his router against prompt injections, finding that rule-based routing blocked fewer than a quarter of attacks and arguing that a guard model is needed instead.

2026-10-15 ~ 2026-10-15 · 2 related posts

• Dev's honest scoreboard: pattern rules catch under 25% of prompt injections, guard model catches most — colinmcnamara · 2026-10-15
• Full talk: pattern lists stop under 25% of prompt injections, guard models catch most — colinmcnamara · 2026-10-15