CSA AI Controls Matrix (AICM)

The practitioner view

Working with CSA AI Controls Matrix (AICM)

1. 01

   What CSA AI Controls Matrix (AICM) is

   CSA control matrix for securing and governing AI systems. It is issued by CLOUD SECURITY ALLIANCE and applies across GLOBAL. Our assessors work against it directly, so what follows is the practitioner view rather than a restatement of the text.

2. 02

   Who it binds

   Applicability turns on what you process and who regulates you, not on your size. If CLOUD SECURITY ALLIANCE sets your rules, or a customer contract names CSA AI Controls Matrix (AICM), you are in scope. We map that boundary first, because an over-drawn scope is the single most expensive mistake in any programme.

3. 03

   What an assessor looks for

   Evidence that the control operated, not that it was written down. Expect sampling across the period, interviews with the control owners, and a request for the artefacts that prove the control ran on the days it was supposed to. Policies without operating evidence are the most common finding we raise.

4. 04

   How to get ready

   Scope it, assess the gap against the current version, remediate in priority order, then collect evidence continuously rather than in the fortnight before the audit. Evidence lag is what slips dates, not the controls themselves.

Free tools

Put a number on it before you call anyone

• FreeGuided readiness estimateA short set of questions returning a readiness score, your top gaps, an indicative cost and a timeline.Open it
• FreeCompliance cost calculatorSize, infrastructure and current readiness turned into an indicative range and timeline.Open it
• FreeAll free toolsCalculators and checkers covering cost, scope, penalty exposure, breach cost and readiness.Open it

Related

More in security frameworks

• NIST Cybersecurity Framework (CSF 2.0)NIST · GLOBALA voluntary, outcome-based framework for managing and reducing cybersecurity risk.
• NIST SP 800-53NIST · GLOBALA comprehensive catalog of security and privacy controls for information systems.
• COBITISACA · GLOBALA governance and management framework for enterprise information and technology.
• CIS ControlsCENTER FOR INTERNET SECURITY · GLOBALA prioritised set of 18 safeguards that stop the most common attacks.
• MITRE ATT&CKMITRE · GLOBALA knowledge base of real-world adversary tactics and techniques.
• SANS / CWE Top 25MITRE / SANS · GLOBALThe most dangerous and common software weaknesses developers must avoid.