AI Governance

A central register of every AI and ML system your organization uses, from ChatGPT to internal models to vendor-embedded AI. Shadow AI is discovered automatically from the OAuth grants in Microsoft 365, Google Workspace, and Okta, then queued for your review.

• Automatic shadow-AI discovery from OAuth grants, plus manual entry
• Name, vendor, purpose, and data access classification
• Sanction workflow: approve or reject with reason, every decision audit-logged
• Risk level, owner assignment, and review tracking with timestamps

Agents are not chatbots. They hold credentials, call tools, and act on their own. The registry records what each agent and MCP server runs as, what it can touch, and how it stops, then flags the gaps regulators are starting to ask about.

• Typed inventory: agents and MCP servers tracked distinctly from AI tools
• Machine identity, tool access, autonomy level, and execution budgets per agent
• Kill-switch and human-approval attestation; unconfirmed agents flagged inline
• Gaps write evidence to OWASP Agentic Top 10 and NIST AI RMF automatically

The agents your own teams build are the ones nobody registers. Radius360 scans connected GitHub repositories for MCP servers, agent frameworks, and AI SDKs, and files what it finds into the registry for review.

• Detects MCP servers and agent frameworks from repository manifests
• Discovered agents enter the same sanction workflow as vendor AI
• Rejected and retired systems stay rejected on re-scan
• Unreviewed discoveries surface as proposals in your inbox

The EU AI Act requires organizations to classify AI systems by risk tier and apply obligations accordingly. Radius360 provides the classification framework with guidance on what each tier requires.

• 4 risk categories: prohibited, high-risk, limited, minimal
• Built-in guidance: which systems fall in which category
• 138 assessable controls from the EU AI Act framework
• Track compliance obligations per system based on classification

The NIST AI Risk Management Framework provides a voluntary, rights-preserving approach to AI risk. Assess your AI systems against the GOVERN, MAP, MEASURE, MANAGE lifecycle.

• 92 controls across 4 core functions
• AI gap assessment with control-by-control suggestions
• Cross-maps to EU AI Act and ISO 42001 controls
• Readiness score per AI framework

The international standard for AI Management Systems. If your organization is pursuing ISO 42001 certification, Radius360 provides the full control set with assessment workflows.

• 43 controls covering AI management system requirements
• Evidence collection mapped to AI-specific controls
• Cross-framework mapping to EU AI Act and NIST AI RMF
• Audit-ready assessment output

Assess your LLM deployments against the OWASP Top 10 for Large Language Model Applications (2025). Covers prompt injection, training data poisoning, model denial of service, and seven more attack categories specific to LLMs.

• 10 risk categories purpose-built for LLM applications
• Control mapping to prompt injection, data leakage, and excessive agency
• Cross-maps to NIST AI RMF and EU AI Act controls
• AI-generated assessment suggestions based on your environment

The first framework designed for autonomous AI agents with tool access, persistent memory, and multi-step reasoning. Covers risks that emerge when AI systems act on their own, not just respond to prompts.

• 10 risk categories for agentic AI: goal hijack, privilege escalation, tool misuse
• Maps to OWASP LLM Top 10 where risks overlap (e.g. prompt injection → goal hijack)
• Control guidance for systems that chain actions without human approval
• Risk scoring per AI system tied to your existing risk register

AI governance isn't one-size-fits-all. Different systems carry different risks. Radius360 lets you apply proportional controls based on each system's classification and data access.

• Risk scoring per AI system tied to your risk register
• High-risk systems flagged for enhanced oversight
• Unapproved AI systems surface in Proactive Insights on your dashboard
• Integration with mitigation task workflow for remediation

A dedicated view showing all AI systems, their risk tiers, approval status, and framework compliance, at a glance.

• KPIs: systems tracked, high-risk count, pending approvals
• Filter by risk level, EU AI Act category, status
• Framework assessment progress for all 5 AI frameworks
• Guidance tab with classification reference and framework explainers