Table of Contents

Reach SOC 2 Compliance in 6 Weeks or Less.

  /

  / AI Governance Maturity Model: 5 Levels Explained

AI Governance Maturity Model: 5 Levels Explained

Most organizations think their AI governance is further along than it is. McKinsey’s 2026 AI Trust Maturity Survey of roughly 500 organizations found an average maturity score of 2.3 out of 4, and only about a third reported level three or higher in strategy, governance, and agentic AI oversight. Adoption is outpacing control, and regulators have noticed.

An AI governance maturity model gives you a way to measure that gap honestly. This guide covers what a maturity model is, the six dimensions it should measure, the five levels most models use, and how to assess your own organization and build a roadmap to the next level.

What Is an AI Governance Maturity Model?

An AI governance maturity model is a structured framework that describes how capable an organization is at governing its AI systems, usually across five progressive levels. The concept borrows directly from the Capability Maturity Model (CMM) that software engineering has used since the early 1990s: define the capability, describe what it looks like at each stage of development, and score yourself against it.

The purpose is diagnosis. A maturity model tells you where governance is strong, where it’s theater, and where it doesn’t exist at all.

How It Differs from General AI Governance Frameworks

Frameworks like the NIST AI Risk Management Framework or ISO/IEC 42001 tell you what good governance contains: policies, risk assessments, accountability structures, monitoring. A maturity model tells you how well you’re doing those things today. The framework is the destination. The maturity model is the odometer.

That distinction matters in practice. Plenty of companies can point to an AI policy document. Far fewer can show that the policy changes what teams actually ship.

Why Enterprises Need a Maturity Model

Three reasons.

  • First, budget: you can’t prioritize governance investment without knowing which dimension lags.
  • Second, accountability: a maturity score gives boards something concrete to track quarter over quarter.
  • Third, regulation: the EU AI Act and frameworks like ISO 42001 assume a functioning management system, and a maturity assessment is the fastest way to find out whether yours would survive scrutiny.

Let Axipro help you build a business continuity plan that's practical, compliant, and audit-ready.

Schedule Your Free Assessment Today

Core Dimensions of an AI Governance Maturity Model

A useful model measures more than policy coverage. Six dimensions show up consistently across the credible models, including the IEEE-USA flexible maturity model built on the NIST AI RMF.

  • Strategy and leadership. Does the organization have a stated position on AI risk, an executive owner (increasingly a Chief AI Officer), and board visibility? Gartner’s 2025 polling found 55% of organizations now have an AI board or dedicated oversight committee, which means nearly half still govern by improvisation.
  • Policies, standards, and accountability. Written policies mapped to regulations, a RACI matrix for AI decisions, and clear escalation paths. Many organizations adapt the three lines of defense model from financial risk: the teams building AI, the risk function overseeing them, and internal audit checking both.
  • Data governance and model lifecycle. Training data lineage, quality controls, and lifecycle management from development through deployment, monitoring, and retirement. This is where AI governance meets MLOps, and where mature organizations maintain an AI register, a live inventory of every model and system in production.
  • Risk, compliance, and ethics. Risk classification of AI systems, impact assessments, bias and fairness testing, and explainability requirements. Banks will recognize the DNA of model risk management under SR 11-7 here.
  • People, skills, and culture. Training, role clarity, and whether people outside the governance team actually understand their obligations.
  • Tools, automation, and monitoring. Drift detection, automated policy checks, audit logging, and dashboards. Governance that lives in spreadsheets caps out around level three.

The 5 Levels of AI Governance Maturity

Level 1: Ad Hoc / Initial

AI use happens without oversight. There’s no inventory, no policy, or a policy nobody follows. Shadow AI is common, and risk surfaces only when something breaks publicly.

Level 2: Developing / Repeatable

Someone has been assigned responsibility. A draft policy exists, a partial inventory exists, and reviews happen for high-profile projects. The practices are repeatable but depend on specific people rather than defined processes.

Level 3: Defined / Structured

Governance is documented, standardized, and applied across the organization. There’s a governance committee, a risk classification scheme, defined lifecycle gates, and mandatory training. Most organizations pursuing ISO 42001 certification are working to reach and formalize this level.

Level 4: Managed / Metrics-Driven

Governance produces numbers. Coverage rates, review cycle times, incident counts, and risk reduction are measured and reported to leadership. Controls are enforced by tooling rather than goodwill, and audits confirm the system works as described.

Level 5: Optimized / Adaptive

Governance improves itself. Monitoring feeds back into policy, controls adapt to new model types (agentic systems being the current test), and the organization anticipates regulatory change rather than reacting to it. Almost nobody is here yet, and that’s fine. Level 5 is a direction, not a deadline.

Insider Note: In assessments, the most common self-scoring error is claiming level 3 on the strength of documents alone. If your policy says every model gets a pre-deployment review and your inventory shows 40 models but your review log shows 6, you’re at level 2. Evidence beats paperwork every time, and auditors check the logs first.

AI Governance Maturity Matrix

The matrix crosses dimensions with levels so you can score each one independently. Organizations are rarely uniform: it’s normal to sit at level 3 on policy and level 1 on monitoring.

For scoring, keep the rubric simple: 1 to 5 per dimension, scored on evidence you could show an auditor, not on intentions. Board-level indicators (does the board see AI risk reporting?) and operational indicators (does every production model have a completed impact assessment?) should be scored separately, because they fail independently.

How to Assess Your Current AI Governance Maturity

Start with a baseline self-assessment. Pull together a cross-functional group covering engineering, legal, risk, security, and the business owners of major AI use cases, and score each dimension against the matrix. Half a day is usually enough for a first pass.

For each dimension, the assessment comes down to a few blunt questions. Do we know every AI system we run? Who approved the last model that went to production, and can we prove it? When did we last test a model for bias or drift? Would our documentation survive an external audit? What happens, step by step, when an AI system misbehaves?

The signs of each level are usually obvious once you look. If the answer to “who owns this?” is a shrug, that’s level 1. If it’s a name, level 2. If it’s a role with a documented mandate, level 3 or better.

For benchmarking, McKinsey’s 2026 data is a reasonable reference point: the average organization sits between levels 2 and 3, with technology and financial services ahead of other sectors. If you’re at level 2, you’re normal. Staying there is the problem.

Pro Tip: Score your Agentic AI

Score your agentic AI use separately from everything else. Autonomous agents that take actions, call tools, and chain decisions break assumptions that traditional model governance relies on, and McKinsey added agentic governance as its own dimension in 2026 for exactly this reason. A level 3 program for predictive models can easily be level 1 for agents.

Aligning the Maturity Model with Established Frameworks

  • NIST AI Risk Management Framework. The AI RMF organizes governance into four functions: Govern, Map, Measure, and Manage. It deliberately avoids prescribing a rigid maturity scale, which is why the IEEE-USA model built a questionnaire and scoring rubric on top of it. If you want a US-anchored, sector-neutral basis for your maturity dimensions, start here.
  • ISO/IEC 42001. Published in December 2023, ISO 42001 is the first certifiable AI management system standard. It follows the same Plan-Do-Check-Act structure as ISO 27001, which makes it a natural extension for organizations that already run an ISMS. In maturity terms, certification roughly demonstrates a defined, audited level 3 with elements of level 4.
  • EU AI Act. The Act’s obligations phase in over several years, and the timeline moved in 2026: under the Digital Omnibus agreed in May and formally endorsed in June 2026, obligations for standalone high-risk systems under Annex III shift to December 2, 2027, and AI embedded in regulated products under Annex I to August 2028. Penalties still reach €35 million or 7% of global turnover for prohibited practices. The extension buys time, but the high-risk obligations still assume exactly the inventory, documentation, and lifecycle controls that levels 3 and 4 describe.
  • OECD and other global standards. The OECD AI Principles, adopted by nearly 50 countries, underpin most national AI policies and give multinationals a common vocabulary. ISO/IEC 23894 adds AI-specific risk management guidance that pairs well with 42001. GDPR continues to apply alongside all of it wherever personal data is involved.

Building a Roadmap to Advance Maturity

Step 1: Establish a baseline. Run the self-assessment above and write the scores down, including the embarrassing ones. The baseline only works if it’s honest.

Step 2: Define a target maturity level. Level 3 across all dimensions is the right target for most enterprises within 12 to 18 months. Regulated industries and anyone deploying high-risk systems under the EU AI Act should aim for level 4 on risk, lifecycle, and monitoring.

Step 3: Prioritize gaps and quick wins. An AI inventory is almost always the first quick win: it’s cheap, fast, and everything else depends on it. Follow with risk classification and a lightweight review gate for new deployments.

Step 4: Run a governance pilot sprint. Pick one business unit or one model class and run the full governance process end to end for 60 to 90 days. A pilot surfaces the friction (unclear ownership, slow reviews, missing evidence) before you scale the pain organization-wide.

Step 5: Scale with automation and tooling. Manual governance doesn’t survive contact with dozens of models. Automate inventory updates, evidence collection, monitoring alerts, and policy checks. Gartner’s 2025 research found organizations investing in third-party AI governance tools were 1.9 times more likely to report high value from generative AI.

Step 6: Quarterly review and continuous improvement. Reassess maturity quarterly, report the trend to the board, and retire controls that create work without reducing risk. Maturity that isn’t re-measured decays.

Let Axipro help you build a business continuity plan that's practical, compliant, and audit-ready.

Schedule Your Free Assessment Today

Metrics and KPIs to Measure Maturity Progress

  • Governance coverage metrics answer “how much of our AI estate is actually governed”: percentage of AI systems in the inventory, percentage risk-classified, percentage with completed impact assessments, and training completion rates.
  • Risk reduction indicators track whether governance changes outcomes: AI incidents per quarter, time to detect and resolve model issues, drift alerts actioned, and bias findings remediated.
  • Operational and efficiency metrics keep governance honest about its own cost: average review cycle time, percentage of reviews meeting SLA, and evidence collection effort per audit.
  • Business outcome and ROI metrics connect governance to value, which is what keeps it funded. Gartner’s 2025 survey found that organizations running regular AI assessments were three times more likely to report high value from generative AI. Done well, governance also speeds delivery, because approvals stop being bespoke investigations.

Common Pitfalls When Implementing a Maturity Model

  1. Treating maturity as a checklist. A maturity model measures capability. An organization that writes twelve policies in a quarter hasn’t advanced a level; it’s produced twelve documents. Advancement means behavior changed and you can prove it.
  2. Ignoring cross-functional ownership. AI governance owned solely by legal produces policies engineers ignore. Owned solely by engineering, it produces monitoring nobody reports upward. The maturity assessment itself should be cross-functional, and so should the operating model that follows.
  3. Underinvesting in monitoring and automation. This is the most common ceiling. Organizations reach level 3 on documentation and stall because every control is manual. Without drift detection, automated evidence collection, and continuous checks, level 4 stays permanently out of reach.

Where to Go from Here

An AI governance maturity model turns a vague obligation into a measurable program: six dimensions, five levels, a baseline score, and a roadmap. The organizations getting this right treat maturity as an operating metric, reassess quarterly, and automate early. With EU AI Act high-risk deadlines now set for late 2027, the window to build toward level 3 deliberately, rather than in a panic, is open but not indefinite.

Frequently Asked Questions

When should an organization adopt an AI governance maturity model?

As soon as AI touches production systems or customer-facing decisions. Earlier is cheaper: retrofitting governance onto dozens of deployed models costs far more than building gates before scale.

A single accountable executive (CAIO, CISO, or chief risk officer depending on structure) with a cross-functional working group doing the scoring. Ownership without cross-functional input produces blind spots; input without a single owner produces stalemate.

Moving from level 1 to 2 can take a quarter. Level 2 to 3 typically takes 6 to 12 months because it requires standardization across teams. Level 3 to 4 depends heavily on tooling investment and often takes another year.

Quarterly for the operational scorecard, annually for a full formal assessment. Anything less frequent and the score stops reflecting reality; AI estates change too fast.

No. The Act requires specific outcomes (risk management, documentation, human oversight, monitoring) rather than any particular maturity model. But a maturity assessment is the most practical way to find out how far you are from those outcomes, and ISO 42001 certification built on a level 3+ program is emerging as the cleanest way to demonstrate readiness.

Axipro Author

Picture of Pedro Dias

Pedro Dias

Pedro has been writing online for over 10 years. With experience in all things programming, cyber security, and compliance, he is our editor-in-chief at Axipro.

Blog Highlights

Explore More Articles

Two compromised versions of LiteLLM sat on PyPI for roughly 40 minutes on the morning of March 24, 2026. That window was enough to capture secrets from around 434,000 CI/CD pipeline runs across nearly 2,500 organizations, including AWS, Samsung, Cisco, Salesforce, Siemens, and Deloitte. In August, researchers at CloudSEK and Hudson Rock confirmed they had obtained the raw exfiltrated data: a 153GB archive containing 433,909 files of environment variables, cloud keys, Kubernetes secrets, and API tokens harvested live from running pipelines, as covered by Help Net Security’s reporting on the credential archive. If LiteLLM runs anywhere in your stack, or you touch any AI proxy infrastructure at all, you need answers to three things: whether you were exposed, what to rotate first, and whether the rotation you did back in March actually held. That last one matters more than it sounds, because “we rotated everything” has already burned at least one very large company. How the Breach Happened The attack didn’t start with LiteLLM. On March 19, 2026, a threat group called TeamPCP compromised the build pipeline of Trivy, a vulnerability scanner half the industry runs, and pushed a poisoned release. LiteLLM’s own CI pipeline ran Trivy, so the poisoned scanner had legitimate read access to the project’s runner environment. The attackers used that to steal LiteLLM’s PyPI publishing tokens and ship two malicious releases of their own: versions 1.82.7 and 1.82.8. KICS and the Telnyx Python SDK got hit in the same campaign. The payload design is the part worth studying. The malicious package dropped a .pth startup hook into site-packages, so the code ran the moment any Python interpreter started on the machine, whether or not anything imported LiteLLM. From there it harvested environment variables, read local credential files like .aws/credentials and .kube/config, tried to move laterally across Kubernetes clusters, and installed a systemd backdoor dressed up as a generic telemetry service. InfoQ’s coverage of the PyPI compromise put downloads of the compromised release above 40,000. For scale, LiteLLM normally gets downloaded around 3 million times a day. The exfiltration had a nasty fallback, too. According to CloudSEK, stolen data was encrypted and sent to a typosquatted domain, and when that failed, the malware created a public repository inside the victim’s own GitHub account and uploaded the loot as a release asset. Some companies were publishing their own secrets to the open internet and had no idea. Worth Knowing: The malicious code only existed in the PyPI artifacts. The GitHub source repository stayed clean the whole time, so a developer reviewing the code on GitHub saw nothing wrong. Source review isn’t artifact verification. If you don’t check that what the registry serves matches the upstream source, this class of attack is invisible to you. How to Check If You Were Exposed Three checks, from quickest to most involved. 1. Confirm whether the compromised versions ever ran The malicious versions went live on PyPI at 10:39 UTC on March 24, 2026 and got quarantined about 40 minutes later. The project’s advice: treat any install from that day before 16:00 UTC as suspect. Search your lockfiles, pip caches, SBOMs, and container image histories for 1.82.7 and 1.82.8. And check your internal artifact mirrors. An Artifactory or Nexus proxy that cached the bad release in March can keep serving it internally long after PyPI pulled it. Keep the .pth mechanism in mind when you scope this. The question isn’t “which applications import LiteLLM,” it’s “which machines had the package installed at all,” because every Python process on an infected machine triggered the payload. 2. Hunt for persistence Rotation is pointless if the attacker still has a foothold. Check developer machines, CI runners, and containers for unauthorized .pth files in site-packages and for suspicious systemd units, especially anything posing as a system telemetry service. And review activity from March 24 onward, not just the 40-minute window. Persistence is there so the access outlives the infection. Pro Tip: Don’t limit the persistence hunt to live machines. Base container images rebuilt in late March may have baked the payload into every image derived from them since. Scan your image registry for the affected LiteLLM versions and for unexpected .pth files, then trace which running workloads came from flagged images. 3. Check whether your secrets are in the dump Hudson Rock has published a domain lookup tool and is running ethical disclosures for affected organizations, and CloudSEK maintains a high-confidence victim list. Use them, but know their limits. Attribution in this dataset is genuinely hard. One dump with a siriusxm.com committer email actually traced, through its self-hosted GitLab endpoints, to AdsWizz, a SiriusXM subsidiary. And a large share of the dumps are generic pipeline configurations with no identifying domain, email, or server name at all. Absence from a victim list is not evidence of absence. If your pipelines ran the compromised versions, assume exposure no matter what a lookup tool tells you. What to Rotate, in What Order The guidance from both research teams is blunt: treat every secret the LiteLLM environment could reach as compromised. That covers secrets on disk, in memory, injected into CI jobs, and anything retrievable through instance metadata services. Work down by blast radius: Priority Credential type Why it comes first 1 Cloud IAM keys (AWS, GCP, Azure) Direct control of infrastructure, data stores, and billing. This is where attackers monetize fastest. 2 GitHub and GitLab PATs, package publishing tokens These let an attacker poison your releases and turn your company into the next link in the supply chain. 3 Kubernetes service account tokens and kubeconfigs Lateral movement across clusters was built into the payload, not a theoretical risk. 4 Database passwords and third-party API keys Dumped in plain text in the archive, often with no attribution, so nobody will warn you they leaked. 5 AI provider API keys Billing abuse, quota theft, and access to whatever data flows through your LLM routing layer. One word matters more than the rest of this article: revoke, don’t just rotate. That

The EU AI Act names recruitment AI as high-risk. Annex III explicitly lists AI systems used for recruitment, candidate selection, and employment decisions, which pulls CV screeners, video interview platforms, and assessment tools into the most demanding compliance regime the Act contains. The original compliance date for these systems was August 2, 2026. In June 2026, the EU’s Digital Omnibus moved the deadline to December 2, 2027, a 16-month extension that has led many HR and talent teams to shelve the topic entirely. That’s a mistake, for two reasons. First, one rule that directly affects recruitment technology is already in force: the ban on emotion recognition in the workplace has applied since February 2, 2025, and it catches features still shipping in some video interview products today. Second, the deferred obligations didn’t shrink. Conformity assessments, human oversight design, bias monitoring, and documentation all still arrive in full, and the practical work of auditing a recruitment stack, renegotiating vendor contracts, and training hiring teams routinely takes a year or more. Here’s what the EU AI Act actually requires of employers and vendors using recruitment tools, on the timeline that now applies. Why Recruitment Tools Are Classified as High-Risk Under the EU AI Act​ Definition of High-Risk AI Systems in Hiring​ The Act takes a list-based approach. Annex III, point 4, designates as high-risk any AI system intended for the recruitment or selection of natural persons, including placing targeted job advertisements, analyzing and filtering applications, and evaluating candidates. The same point covers AI used for decisions on promotion, termination, task allocation, and monitoring of workers, so the classification follows the tool through the entire employment lifecycle, not just the hiring funnel. The reasoning is straightforward: hiring decisions shape access to livelihoods, and algorithmic discrimination in hiring is well documented. The European Commission’s regulatory framework for AI treats employment as one of the areas where an AI error or bias causes serious harm to fundamental rights. That’s the test for the high-risk tier. Types of Recruitment Tools Affected In practice, the high-risk classification captures most of the modern recruitment stack: CV and resume screeners that rank or filter applicants, video interview platforms that score responses or delivery, psychometric and skills assessment tools that produce scores feeding a hiring decision, sourcing and matching algorithms that decide which candidates a recruiter sees, and programmatic job ad targeting systems that determine who sees a vacancy at all. If the system’s output materially influences who advances and who does not, assume high-risk until proven otherwise. Important: Emotion recognition is not high-risk in the workplace. It is prohibited. Article 5 bans AI systems that infer emotions of people in the workplace (outside narrow medical and safety cases), and that ban has applied since February 2025 with the Act’s top penalty tier attached. If your video interview vendor markets “engagement scoring” or “sentiment analysis” of candidates, that feature needs to be switched off for EU hiring now, not in 2027. Recruitment Tools That May Fall Outside High-Risk Classification Not everything in the HR stack qualifies. The Act carves out systems performing narrow procedural tasks that do not materially influence decision outcomes. An applicant tracking system that stores applications, schedules interviews, and sends templated emails is a database with a workflow, not a high-risk AI system. The same goes for tools that transcribe interviews without scoring them, deduplicate candidate records, or generate first drafts of job descriptions for a human to edit. The line is decision influence: the moment a tool ranks, scores, filters, or recommends candidates, it crosses into Annex III territory. Deployers who rely on an exemption must be able to document that assessment, so “we decided it doesn’t count” needs to exist on paper. Extraterritorial Scope: Which Employers Are Covered The Act applies to providers placing AI systems on the EU market and to deployers established in the EU, but it also reaches further: it covers providers and deployers located outside the EU where the output of the system is used in the EU. For recruitment, the consequence is blunt. A US or UK company with no EU entity that uses an AI screener to filter applicants for roles based in Berlin or Dublin, or that screens candidates located in the EU, is using the system’s output in the Union. Brexit doesn’t move UK employers out of scope when they hire into or from the EU. Providers vs. Deployers of Recruitment AI Tools The Act splits obligations between the provider (the vendor that develops the tool and places it on the market) and the deployer (the employer using it). Most employers are deployers, and deployer obligations are lighter but real. One common trap: an employer that substantially modifies a high-risk system, or puts its own name on it, can be reclassified as a provider and inherit the full provider stack. Heavy customization of a screening model, or fine-tuning it on your own hiring data, can be enough to trigger this. Key Obligations for Employers Using AI Recruitment Tools Human Oversight in Automated Hiring Decisions Deployers must assign oversight of the system to people with the competence, training, and authority to intervene. That last word matters. A recruiter who rubber-stamps whatever the ranking algorithm produces, because nobody has time to review 800 rejected CVs, doesn’t count as oversight. Regulators and courts will look at whether the human could genuinely override the system and whether they ever did. Designing review checkpoints where a person can meaningfully change the outcome, and logging when they do, is the core of compliant deployment. Transparency Requirements Toward Candidates Employers must inform workers and their representatives before putting a high-risk AI system into use at work, and candidates subjected to such a system must be told it is being used. In countries with works councils, such as Germany, this obligation lands on top of existing co-determination rights, so employee representatives may need to be consulted before the tool goes live rather than just told afterward. Burying an AI disclosure in a privacy policy paragraph is unlikely to survive scrutiny.

A green dashboard is not an audit opinion. Compliance automation platforms like Vanta, Drata, Secureframe, and Hyperproof have made SOC 2 readiness faster and cheaper, but every audit cycle produces the same pattern: controls that sat at “passing” for months come back from the auditor with exceptions or requests for re-testing. The four controls below account for a disproportionate share of those rejections, and they all fail for the same underlying reason. The tool confirmed that evidence exists. The auditor tested whether the control actually operated. This article walks through each of the four: what auditors reject, why, and how to fix the evidence before fieldwork starts. Why Compliance Tools Show “Passing” But Auditors Still Reject Controls​ The Gap Between Automated Checks and Auditor Judgment Compliance platforms run continuous control monitoring: API calls that check whether a configuration exists, a document is uploaded, or a task is marked done. That’s real value. It catches drift, keeps evidence in one place, and saves weeks of screenshot collection. An audit is a different exercise. A SOC 2 examination is an attestation performed by a CPA firm under AICPA standards, and the auditor’s job is to form an independent opinion on whether your controls met the Trust Services Criteria. That opinion rests on professional judgment, not on whether an API integration returned a 200 response. What “Passing” Actually Means in Your Compliance Dashboard​ When a control shows “passing,” the platform is telling you one narrow thing: at the moment of the last scan, an automated test found the artifact or setting it was programmed to look for: MFA enforced in the identity provider, a policy document uploaded, a training campaign sitting at 100%. The test says nothing about whether the underlying process ran the way your control narrative claims it did, or whether it ran that way across the whole audit period. How Auditors Evaluate Controls Beyond the Checkbox Auditors test two dimensions. Design effectiveness asks whether the control, as described, would meet the criterion if it worked as intended. Operating effectiveness, the core of a SOC 2 Type 2 report, asks whether it actually did throughout the audit period. To answer that, the auditor pulls a population (every access review, every change, every new hire in the period), selects a sample, and inspects the evidence item by item. A dashboard status feeds into that process. It doesn’t replace it. Insider Note: Auditors increasingly ask for evidence outside the compliance platform precisely because they know what the platform auto-collects. If every artifact you produce comes from the same tool export, expect the auditor to independently pull the population from the source system and compare. Discrepancies between the two are one of the fastest routes to an exception. Control #1: Access Reviews That Automation Marks Complete but Auditors Reject Why Auditors Reject Automated Access Review Evidence​ User access reviews sit under the logical access criteria (CC6.1 through CC6.3), and they are the single most common source of audit exceptions we see. The typical failure: the platform generated a user list, someone clicked “complete,” and the dashboard turned green. The auditor then asks a simple question the evidence can’t answer: what did the reviewer actually decide? The Missing Element: Documented Reviewer Judgment​ An access review is a judgment control. Someone with knowledge of the system must look at each account and confirm the access is still appropriate for the person’s role. A timestamped task closure proves the task was closed. It doesn’t prove anyone assessed anything, and an “approve all” review completed in ninety seconds gets exactly the skepticism it deserves. What Auditors Actually Want to See in Access Review Evidence Auditors look for four things: The full population of accounts at the time of review (including service accounts and admin roles), Evidence of who reviewed it and when, explicit dispositions per account or group (retain, modify, revoke), and Proof that flagged access was actually removed. That last item, the deprovisioning ticket showing revocation within a defined window, is the piece most companies can’t produce. How to Fix Your Access Review Control Before the Audit​ Assign a named control owner per in-scope system, run reviews quarterly, and require reviewers to record a disposition for every line, not a blanket approval. When access is revoked, link the removal ticket to the review record. If a quarter was missed, don’t backfill it. Document it honestly and show the remediation, because auditors treat fabricated retroactive evidence far more severely than a disclosed gap. Control #2: Change Management Approvals That Pass Automated Scans​ Why Ticket Closure Isn’t Proof of Approval​ Change management (CC8.1) automation typically verifies that production changes link to a ticket and the ticket is closed. Auditors test something stricter: that each sampled change was approved by an authorized person before deployment. An approval added after the merge, or a ticket closed by the same engineer who wrote the code, fails that test even though every automated check came back green. The Segregation of Duties Problem Automation Misses Segregation of duties is the requirement that no single person can develop, approve, and deploy the same change. NIST’s SP 800-53 control catalog treats it as a foundational access control principle, and SOC 2 auditors apply the same logic. Small engineering teams trip on this constantly. Self-approved pull requests, admins who can bypass branch protection, direct pushes to main: a scanner sees “changes with tickets” while an auditor sees SoD violations. Emergency Changes and Retroactive Approvals: Common Rejection Triggers​ Every audit period contains hotfixes. Auditors don’t reject emergency changes. They reject emergency changes with no documented post-hoc review. If your policy says urgent changes get retroactive approval within two business days, the auditor will sample your emergency changes and check exactly that. No policy, or a policy nobody followed, produces an exception. Rebuilding Change Management Evidence Auditors Will Accept​ Enforce the control technically: branch protection requiring at least one independent reviewer, no admin bypass, and deploy pipelines that only run from protected branches. Then write the emergency change procedure down and generate the review artifact every time it fires.