CalypsoAI

CalypsoAI is now part of F5 and is offered as F5 AI Guardrails, a security product for AI models, applications and agents. It sits between users, applications and models to enforce policy at runtime, defending against prompt injection, jailbreaks, data exfiltration and harmful outputs. The listed address now points to the F5 product page. Key capabilities include model agnostic policy enforcement across public and proprietary models, and custom guardrails that can be created through a natural language interface for a given use case, region or industry. Day one compliance controls address personal data leakage, EU AI Act alignment, PCI and PHI protection. For agents, it enforces limits on agent actions and tool use to prevent excessive agency and privilege escalation. Content moderation aligns outputs with enterprise definitions of biased, toxic or harmful content, and audit ready logging traces each enforcement action with guardrail attribution. A continuously updated attack library adds over 10,000 attack patterns monthly, and the CASI and ARS leaderboards compare the security risk profiles of popular models. It is aimed at enterprise security, governance and platform teams deploying generative AI across cloud and hybrid environments. F5 does not publish pricing for AI Guardrails on this page. A demo is available and purchase is handled through the vendor, so cost is not stated.