New Open Source Framework 'b3' Enhances LLM Security for AI Agents

New Open Source Framework 'b3' Enhances LLM Security for AI Agents

First seen 2 Nov 2025, 16:14 UTC •Infosecurity-MagazineSecuritybrief •21.6

Article Content

Browse articles
ThreatCluster

Check Point, Lakera, and the UK AI Security Institute have launched b3, an open-source framework aimed at testing the security of large language models (LLMs) used in AI agents. The b3 benchmark introduces a methodology focusing on 'threat snapshots' to identify vulnerabilities in AI workflows, allowing developers to improve the security posture of LLMs that power AI agents.

How this analysis works

Extracted Entities

1 / 2