November 12, 2026 · Google CloudFeature
Model Armor filter v4 improves prompt injection and jailbreak detection
Filter version v4 under the Latest alias includes an updated prompt
injection and jailbreak detection
filter with the following improvements:
- Higher recall for indirect prompt injection (IPI) and tool calls.
- Higher recall for conversational inputs at a lower false positive rate.
- Higher recall for jailbreak attacks.
For more information, see Filter version history.
Text from the Google Cloud release notes, licensed under CC BY 4.0. Source: Google Cloud release notes.
Related updates
Model Armor improves prompt injection and jailbreak protection for Workspace data
Public Preview: Agent mitigations and guardrail customization · Azure
Data Loss Prevention - New DLP topic based detection entries for AI prompt protection · Cloudflare
WAF - Updated unsafe topic detection for AI Security for Apps · Cloudflare