Probative

This safety assurance tool systematically tests artificial intelligence systems for vulnerabilities by conducting extensive, multi-faceted attack probes across eleven identified threat classes, including jailbreaks, prompt injections, data exfiltration, bias, hallucination, toxicity, refusal, and drift. It automates continuous re-verification through scheduled scans at a single endpoint—such as an API or chat interface—producing cryptographically signed records of each assessment, enabling transparent, verifiable audit trails. Designed for AI developers, security teams, and risk managers, it helps safeguard models against malicious exploitation, ensures compliance with governance standards, and builds trust by proactively identifying and mitigating operational and security risks in AI deployments.

Probative can be found in API Security categories.