CalypsoAI
10.62.4 commercialCalypsoAI is a security platform that detects and prevents AI-generated content risks, misuse, and vulnerabilities in applications.
Summary
CalypsoAI is a commercial ai-security platform that lets organizations manage and audit the use of AI models and agents across their teams, run as a hosted service accessed through an API rather than installed locally. Its endpoints cover role-based access control, project and provider management, scanner packages, prompt templates, and campaign-based testing runs, alongside audit logging and chat history, pointing to a tool built for security and compliance teams overseeing AI deployments rather than for developers building models themselves. It sits in the AI-governance and red-teaming space, combining access control with structured scanning and audit trails rather than leaving those as separate tools. With 12 releases in the tracked window and the last one 24 days ago, it is under active, frequent development.
CalypsoAI is a security platform that detects and prevents AI-generated content risks, misuse, and vulnerabilities in applications.
What CalypsoAI answers
Which parts of an AI deployment can this restrict access to, versus only observe?
roles and permissions govern projects, providers, and users, but chats and audit records are readable and deletable rather than access-scoped to a finer grain
How would a security team reconstruct what happened after an incident?
audit events and chat history are queried individually by ID or listed in full, giving a record of activity tied to specific users and sessions
Can multiple teams or business units keep their AI usage separate?
providers, scanners, and members are all scoped under a project, so access and testing setups for one group do not carry over to another
What happens to a scanning campaign once it has run?
campaign runs and schedules exist as records that can be listed, fetched, or removed independently of the campaign definition that created them
Do I need to rebuild role definitions from scratch?
a default role is exposed alongside custom ones, so a starting set of permissions exists before any team-specific roles are added
Is offboarding a user a single action or several?
a user's account, multifactor setup, and role assignments are removed through separate calls, so cleanup is a sequence rather than one deletion
Release history
- 10.62.4
CalypsoAI now publishes an API — 141 endpoints across 28 areas: Projects, Users, Scanners, …
- ›Projects (15 endpoints) — create, read, update, delete
- ›Users (12 endpoints) — create, read, update, delete
- ›Scanners (10 endpoints) — create, read, update, delete
- ›Prompts (9 endpoints) — create, read, update, delete
- ›Providers (9 endpoints) — create, read, update, delete
+4 moreshow less
- ›Auth (8 endpoints) — create, read, update, delete
- ›Branding (8 endpoints) — create, read, update, delete
- ›Chats (6 endpoints) — create, read, update, delete
- ›20 more areas: Datasets, Scanner Packages, Campaign Runs, Campaign Schedules, Campaigns, Org, Scans, Secrets, Openai, Refusal Catalogs, Tokens, Admin, Audit, Files, Agent Sessions, Anthropic, Generations, Google, License, Project