The intelligence layer for AI security

AIRiskDB is SuperAlign's continuously updated intelligence database, analysing thousands of AI assets and their associated threats to provide actionable risk insights for security and governance workflows.

24,000+
Apps indexed
12,000+
MCP servers analysed
8+
Types of assets covered
Coverage

Every AI asset, one risk language.

AIRiskDB treats AI tools as a broad asset class — from SaaS copilots and chatbots to MCP servers, browser extensions, IDE plugins, and foundational models — and normalizes them into a single, comparable risk model.

Signals AIRiskDB analyzes

  • Vendor privacy policies, and data collection and sharing disclosures.
  • Terms of service and usage rights, including training‑data and output clauses.
  • Security posture: encryption, access controls, incident history, and operational safeguards.
  • Compliance evidence across regimes like SOC 2, ISO 27001, and sector‑specific obligations, where disclosed.
  • Known AI‑specific vulnerabilities, incidents, and supply‑chain risks from research and public reporting.
Risk Taxonomy

Unified AI risk taxonomy, mapped to the frameworks you already use.

AIRiskDB is built on SuperAlign's AI risk taxonomy, which synthesizes insights from OWASP LLM Top 10, MITRE ATLAS, MIT AI Risk Repository, IBM AI Risk Atlas, and Cisco AI safety taxonomies. It explicitly ties every risk category back to widely adopted frameworks like NIST AI RMF, the EU AI Act, and ISO 42001.

Core risk domains

01

Data security & privacy

Data flows, retention, exposure, and privacy risk across AI tools and vendors.

02

Model safety & robustness

Alignment failures, hallucinations, adversarial inputs, and reliability gaps.

03

Security attacks & supply chain

Prompt injection, tool poisoning, model supply chain compromise, and agent exploitation.

04

Human–machine interaction & governance

Oversight gaps, automation bias, accountability, and human control mechanisms.

05

Regulatory & compliance

Obligations under NIST AI RMF, EU AI Act, ISO 42001, and sector-specific requirements.

Get started

The shared risk brain for SuperAlign Radar and Surface.

A shared intelligence layer and a consistent AI risk language across security and governance workflows.

Request a demo →

Also in the SuperAlign suite