AI Security Certifications Compared: AIGP, AAISM, SecAI+

AI security and governance credentials have multiplied in two years, and most people choosing one cannot tell from the marketing which is a management credential, which is an audit credential and which tests hands-on security skill. This guide puts the main options side by side: the IAPP Artificial Intelligence Governance Professional (AIGP), ISACA Advanced in AI Security Management (AAISM), ISACA Advanced in AI Audit (AAIA), CompTIA SecAI+, the Cloud Security Alliance Trusted AI Safety Expert (TAISE) and the AWS Certified AI Practitioner as an example of a vendor credential. Every exam fact here comes from the certifying body's own pages, checked on 27 October 2026; fees and formats change, so confirm before you book. It is written for security engineers, GRC and audit staff, and managers who fund training budgets and want to know what a credential actually proves.

Key takeaways

Decide the job first, then the credential. AIGP is governance and law, AAISM is security management, AAIA is audit, SecAI+ is practitioner security, TAISE is a broad training-led credential and vendor certificates prove platform knowledge.
The ISACA pair are add-ons, not entry points. AAISM requires an active CISM or CISSP and AAIA requires CISA or another qualified designation, so they extend an existing career rather than start one.
SecAI+ is the only one built around securing and using AI hands-on. CompTIA weights "Securing AI systems" at 40% of the CY0-001 exam and includes performance-based questions.
Total cost is more than the exam fee. Add application fees, annual maintenance, CPE time, training bundles and retakes when you compare; TAISE, for example, is sold only as training plus exam.
No credential replaces evidence of work. Pair any of these with a threat model, red team report or AI risk assessment you have produced, because hiring managers ask for both.

The credentials at a glance

The table below uses only what each certifying body publishes. Where a body does not state a figure on its credential page, the cell says so rather than filling the gap from third-party sites.

Credential	Body	Focus	Prerequisite	Exam format (as published)	Exam fee
AIGP	IAPP	AI governance, law, standards, development and deployment oversight	None stated on the credential page	100 questions, 2.75 hours plus a 15-minute break, Pearson VUE centre or OnVue online	US$649 member, US$799 non-member
AAISM	ISACA	AI security management: governance, controls, risk	Active CISM or CISSP	Computer-based, PSI centre or remote proctored; question count and duration not stated on the credential page	US$459 member, US$599 non-member, plus US$50 application fee
AAIA	ISACA	Auditing AI systems	CISA, or a listed designation (for example CIA, US CPA, ACCA, ICAEW ACA) held by someone in an IT audit or advisory role	Computer-based, PSI centre or remote proctored; remote proctoring not available to residents of India, Mainland China and Hong Kong	US$459 member, US$599 non-member, plus US$50 application fee
SecAI+ (CY0-001)	CompTIA	Securing AI systems and using AI in security operations	None required; CompTIA recommends 3-4 years in IT and 2+ years hands-on cybersecurity, with Security+, CySA+, PenTest+ or equivalent	Maximum 60 questions, multiple choice and performance-based, 60 minutes, pass mark 600 on a 100-900 scale	Not stated on the certification page
TAISE	Cloud Security Alliance	Broad AI safety and security across ten modules	None; CSA recommends basic AI, cloud and security familiarity and suggests CCSK	60 multiple-choice questions (6 per module), 120 minutes, 80% to pass	US$795 bundle including training and two exam attempts valid for two years
AWS Certified AI Practitioner (AIF-C01)	AWS	Foundational AI, ML and generative AI on AWS	None; aimed at people familiar with AWS AI/ML services who do not necessarily build them	65 questions, 90 minutes	US$100

For the deeper detail on each, see the IAPP AIGP (AI Governance Professional) Guide, the ISACA AAISM and AAIA AI Certifications Guide and the CompTIA SecAI+ Guide. This page compares them; it does not repeat their study advice.

What each exam actually covers

Domain names are the clearest signal of what a credential tests. Weightings matter more than titles: a credential that mentions "security" in one of four domains is a governance credential with a security chapter, not a security credential.

Credential	Published domains	Weighting (as published)
AIGP (Body of Knowledge v2.1, effective 2 February 2026)	I Understanding the foundations of AI governance; II Understanding how laws, standards and frameworks apply to AI; III Understanding how to govern AI development; IV Understanding how to govern AI deployment and use	Scored question ranges: I 16-20, II 19-23, III 21-25, IV 21-25
AAISM	AI Governance and Program Management; AI Technologies and Controls; AI Risk Management	Not stated on the credential page
AAIA	AI Governance and Risk; AI Operations; AI Auditing Tools and Techniques	Not stated on the credential page
SecAI+ (V1)	Basic AI concepts related to cybersecurity; Securing AI systems; AI-assisted security; AI governance, risk, and compliance	17%, 40%, 24%, 19%
TAISE	Ten modules from Introduction to AI through Generative AI Architecture and Design, AI Model Lifecycle and Threat Taxonomy, Model Governance, Risk Management, and Compliance, Cloud and AI Security, and Data Security and Privacy AI Systems	Equal: 6 questions per module

Governance depth: AIGP gives the most weight to law and frameworks (domain II alone carries 19-23 scored questions, about a quarter of the exam), which suits people who must read the EU AI Act, ISO/IEC 42001 and the NIST AI RMF against their own programme. See the EU AI Act Compliance Guide, the ISO/IEC 42001 Guide and the NIST AI RMF Guide.
Technical depth: SecAI+ is the only credential here whose largest domain is securing AI systems, and the only one with performance-based items. It is the closest fit for engineers who defend model endpoints, retrieval pipelines and agents.
Management depth: AAISM assumes the candidate already runs or advises a security programme (that is why it requires CISM or CISSP) and adds AI technologies, controls and risk on top.
Audit depth: AAIA is the only credential with a domain dedicated to audit tools and techniques for AI, which matters for internal audit and assurance teams asked to test AI controls.

Matching credentials to roles

Role	First choice	Useful second	Why
Security engineer or architect securing AI applications and agents	SecAI+	A cloud vendor AI credential for the platform you run	Hands-on weighting; vendor credential covers the services you configure
CISO, security manager, head of security architecture	AAISM (if you hold CISM or CISSP)	AIGP	Management and controls focus; AIGP adds the legal layer
AI governance lead, privacy professional, responsible AI officer	AIGP	ISO/IEC 42001 lead implementer	Laws and frameworks are the core of the job; see the ISO 42001 Lead Auditor and Lead Implementer Guide
Internal or external IT auditor	AAIA (with CISA)	ISO/IEC 42001 lead auditor	Audit techniques for AI; certification-audit route for management systems
Cloud security generalist new to AI	TAISE or AWS Certified AI Practitioner	SecAI+ once experienced	Broad foundation without prerequisites
Penetration tester or red teamer	SecAI+	None of these is a red team practical; build portfolio evidence	See the AI Red Teaming Guide

Small teams versus large organisations

Small teams: one person often covers security and governance. A practitioner credential (SecAI+) plus working knowledge of one framework is usually better value than two governance credentials.
Large organisations: split the investment by function. Engineering takes SecAI+ or vendor credentials, the second line takes AIGP or AAISM, internal audit takes AAIA. Align this with the three lines model your risk function already uses, as described in the AI Model Risk Management Guide.
Regulated sectors: where supervisors expect named accountable owners for AI risk, governance credentials help demonstrate competence, but supervisors judge the control evidence, not certificates.

Cost, maintenance and the real time commitment

Compare five-year cost, not exam price. The figures below are published by each body; where nothing is published on the pages checked, plan for it and confirm directly.

Cost element	AIGP	AAISM / AAIA	SecAI+	TAISE	AWS AI Practitioner
Exam	US$649 / US$799	US$459 / US$599 plus US$50 application	Check CompTIA store	US$795 bundle	US$100
Booking window	Complete within one year of purchase	Eligibility six months from registration; AAIA candidates must apply within five years of passing	Check voucher terms	Two attempts valid for two years	Check AWS pricing page
Maintenance	See IAPP candidate handbook	AAISM: 10 CPE hours a year and 30 AAISM-related CPE hours over three years; annual fee US$20 member, US$35 non-member, one fee covers multiple ISACA Advanced credentials	CompTIA estimates retirement of the V1 exam about three years after the 17 February 2026 launch	Not stated on the page checked	Valid three years; recertify by passing the latest exam or the AWS Certified Machine Learning Engineer Associate

Hidden cost 1, the prerequisite: AAISM and AAIA only make sense if you already hold (and pay to maintain) CISM, CISSP or CISA. Count that maintenance if you would otherwise let it lapse.
Hidden cost 2, the training bundle: TAISE has no exam-only option, so the price includes training whether you need it or not.
Hidden cost 3, exam churn: AI exams are refreshed quickly. AIGP is already on Body of Knowledge v2.1 and SecAI+ has a published retirement estimate; budget time to refresh study material.

How to evaluate a credential before you fund it

Read the official blueprint, not a summary. Download the body of knowledge or exam objectives and highlight every objective you will use in the next 12 months. Below about half, it is the wrong credential for the role.
Check the issuer and accreditation. Ask whether the credential is a certification (exam plus maintenance and ethics code) or a certificate of course completion. Check whether the body states accreditation such as ISO/IEC 17024 on its own pages; do not rely on training providers' claims.
Check prerequisites and timing. AAISM and AAIA gate on existing ISACA or ISC2 credentials; SecAI+ recommends prior CompTIA-level skills. Plan the sequence before booking.
Price the five-year total. Exam, application, membership if it lowers the fee, annual maintenance, CPE hours valued at staff time, retakes.
Agree the evidence it must produce. For each funded credential, agree one work product: an AI threat model, an AI risk assessment, an audit programme or a red team report. See the AI Agent Threat Model Template.
Review after the first exam cycle. Ask candidates which objectives were relevant and drop credentials that did not change how they work.

Where identity fits

Most AI security incidents that reach the news are credential and permission failures: API keys in code, over-scoped service accounts behind model endpoints and agents acting with a human's full rights. None of the credentials here has a domain named for identity, so check how much of "Securing AI systems" (SecAI+) or "AI Technologies and Controls" (AAISM) touches authentication, authorisation and secrets in practice, and fill the gap with identity training. See the Agent Identity and Access Guide and the Agentic AI Identity Guide. The Google API Keys Exposure 2026 case shows why: keys long treated as public became sensitive once an AI service was enabled on the same project, a risk that sits between AI security and identity hygiene.

Mapping credentials to the frameworks you must work with

This table maps published domain titles to the frameworks practitioners are asked about. It shows where a domain plainly addresses the topic; it does not claim that a body tests any specific clause.

Framework or obligation	Where it sits	Credential domains that address the area
EU AI Act, national AI laws	Regulation	AIGP domain II (laws, standards and frameworks); SecAI+ AI governance, risk, and compliance
ISO/IEC 42001 AI management system	Certifiable management system standard	AIGP domain II; AAISM AI Governance and Program Management; AAIA AI Governance and Risk
NIST AI RMF (Govern, Map, Measure, Manage)	Voluntary risk framework	AAISM AI Risk Management; AIGP domains III and IV
NIST CSF 2.0 GV.RR (roles, responsibilities and authorities) and PR.AT (awareness and training)	Workforce and training outcomes	Any credential, as evidence of role competence
NIST SP 800-53 Rev. 5 AT-3 Role-Based Training	Control for training by role	Map each funded credential to the role it trains
OWASP Top 10 for LLM Applications 2025, MITRE ATLAS	Threat taxonomies	SecAI+ Securing AI systems; TAISE AI Model Lifecycle and Threat Taxonomy
Internal audit of AI controls	Assurance	AAIA AI Auditing Tools and Techniques

Measuring whether the training budget works

Coverage: percentage of staff in AI-facing roles (model platform, AI governance, audit of AI) holding a credential matched to their role in the table above.
Work products: number of AI threat models, risk assessments or audit programmes produced by certified staff in the year after passing.
Findings quality: share of AI security findings (from reviews, tests or audits) that name a specific control, owner and framework reference.
Maintenance health: lapsed credentials and missed CPE deadlines per year; ISACA Advanced credentials need annual CPE and fees by 1 January.
Relevance score: candidates' own rating of how many blueprint objectives they used within six months.

Common failure patterns

Buying the governance credential for the engineer. An engineer with AIGP can explain the EU AI Act but may still ship an agent with a long-lived API key. Match the credential to the work.
Ignoring the prerequisite chain. Teams book AAISM or AAIA for staff who do not hold CISM, CISSP or CISA, then discover the credential cannot be awarded.
Trusting third-party exam summaries. Training sites often publish outdated question counts or fees. Use the certifying body's blueprint and store page, and record the date you checked.
Treating a course certificate as a certification. A completion certificate with no maintenance or ethics code proves attendance, not competence.
No identity coverage. AI programmes staffed with certified people still fail on secrets, service principals and agent permissions when nobody owns machine identity.
One credential for the whole team. Everyone sits the same exam, and the audit, governance and engineering gaps stay open.

Practitioner checklist

Write down the AI work each person will do in the next 12 months before choosing a credential.
Download the official blueprint or body of knowledge and check objective relevance.
Confirm prerequisites (CISM or CISSP for AAISM; CISA or listed designations for AAIA).
Price exam, application, maintenance, CPE time and retakes over five years.
Record the date you checked each fee and format, and recheck before booking.
Agree one concrete work product per funded credential.
Add identity, secrets and agent permission training where the blueprint is thin.
Track CPE and maintenance deadlines centrally.
Review the credential mix every year as exams are revised.

AI Security Certifications Compared: AIGP, AAISM, AAIA, SecAI+ and TAISE side by side