RESILIAM INSIGHTS

For many organizations today, artificial intelligence has shifted from a value-add to a business necessity. Whether organizations are deploying generative AI tools, using AI to assist in decision-making, or embedding machine learning into core products and services, AI is increasingly business-critical. With that shift comes a familiar challenge: how do organizations govern AI in a way that is responsible, defensible, and scalable over time?

ISO/IEC 42001:2023 (“ISO 42001”), the international standard for Artificial Intelligence Management Systems (“AIMS”), offers a practical answer. Rather than focusing on any single technology or regulation, ISO 42001 provides a structured, certifiable framework for governing AI across its lifecycle. This article provides a practical overview of ISO 42001, explains why it matters, and highlights how organizations can use it to strengthen trust, manage risk, and future-proof their approach to AI.

What Is ISO 42001?

ISO 42001 is the first international management system standard dedicated specifically to artificial intelligence. Published in late 2023, it defines requirements and guidance for establishing, implementing, maintaining, and continually improving an Artificial Intelligence Management System.

Like other ISO management system standards (such as ISO 27001 for information security or ISO 9001 for quality), ISO 42001 follows a familiar structure. The standard includes requirements for governance, leadership, performance evaluation, and continual improvement, all within the context of the organization’s in-scope AI technologies and supporting resources. This design allows ISO 42001 to function either as a standalone management system or to integrate cleanly with existing management systems, extending established governance practices into the AI domain rather than starting from scratch. For organizations already operating mature security, privacy, or risk management programs, this alignment significantly lowers the barrier to adoption.

Why AI Governance Requires a Management System

AI introduces risks and impacts that differ meaningfully from traditional IT systems. These include, for example:

• Inaccurate, misleading, or biased outputs; 
• Enhanced privacy, security, and regulatory concerns; and
• Over-reliance on AI outputs without sufficient human oversight.

Many organizations attempt to address AI risk through siloed efforts across IT, legal, security, and privacy teams. While well-intentioned, these fragmented efforts often fail to scale and become increasingly difficult to coordinate as AI use expands. ISO 42001 offers a collaborative approach: rather than treating AI governance as a collection of disconnected controls, the standard establishes an interconnected management system built around consistency, accountability, and continual improvement.

Scope and Applicability Across Industries

Among ISO 42001’s greatest merits is its flexibility; the standard is designed to be applicable to organizations of all sizes and across all sectors. Organizations define the scope of their AIMS based on how they interact with AI systems.

ISO 42001 recognizes several common roles, including:

• AI developers, who design and build AI systems;
• AI providers, who offer AI-enabled products or services; and
• AI users, who deploy or rely on AI systems developed by others.

Most organizations will fall into one or more of these categories. A company using third-party generative AI tools, for example, may primarily be an AI user, while also developing internal models for specific business functions. ISO 42001 accommodates this reality by allowing organizations to tailor requirements and controls to their actual use cases.

Core Components of an AI Management System

At its core, ISO 42001 requires organizations to establish governance processes that span the full AI lifecycle. Key components include the following.

AI Policies and Acceptable Use

Organizations must document policies governing the development or use of AI systems. These policies typically address topics such as:

• Approved AI systems and use cases;
• Expectations for human oversight and professional judgment; and
• Processes for reporting incidents or concerns.

Well-designed AI policies ensure discipline without stifling innovation, helping users understand both the capabilities and limitations of AI tools.

Governance Structure and Accountability

ISO 42001 places strong emphasis on organizational accountability. This includes defining roles and responsibilities for AI governance, establishing escalation paths for AI-related concerns, and ensuring leadership oversight of AI risks and objectives.

Clear accountability helps avoid common missteps such as “shadow AI” deployments, inconsistent decision-making, or unclear ownership when issues arise.

AI System Impact Assessments and AI Risk Assessments

ISO 42001 introduces a structured approach to understanding and managing AI-related risks through the use of both AI system impact assessments and AI risk assessments. While related, these assessments serve distinct but complementary purposes within an organization’s AIMS.

AI system impact assessments evaluate the potential positive and negative impacts of an AI system’s deployment, intended use, and foreseeable misuse on individuals, groups, and society. These assessments help organizations understand how AI systems may affect their ability to achieve their AIMS objectives (e.g., accountability, security, privacy, and transparency), as well as provide important context for evaluating downstream organizational risks.

AI risk assessments build on this foundation by examining potential events and outcomes related to the development or use of AI systems and the consequences those outcomes may have for the organization. In contrast to traditional IT risk assessments, ISO 42001 requires organizations to consider both negative risks that must be mitigated and positive risks that may present opportunities for improvement and increased effectiveness. Organizations may also look to supporting guidance, such as ISO/IEC 23894, when designing and executing AI risk assessment processes.

Together, AI system impact assessments and AI risk assessments support informed decision-making and continual improvement across the AI system lifecycle, helping organizations manage risk while demonstrating responsible and trustworthy use of AI.

Lifecycle Management and Monitoring

ISO 42001 requires organizations to define criteria and controls across the AI system lifecycle, including requirements and specification, verification and validation, deployment, and ongoing operation and monitoring. Some controls apply at the time of AI system development or procurement, while others (such as performance monitoring and change management) become critical once systems are deployed and in use.

This lifecycle-based approach closely aligns with robust project management practices, helping organizations apply appropriate governance at each stage rather than relying on one-time reviews. The specific controls implemented will vary depending on whether an organization is developing AI systems or using those provided by third parties, but in all cases, continuous oversight is essential to maintaining trust, effectiveness, and alignment with organizational objectives.

Third-Party Management

Few organizations build or operate AI systems entirely on their own. As a result, ISO 42001 includes requirements for managing third parties involved in the AI lifecycle, including AI vendors, service providers, and, in some cases, clients or business partners that rely on or are affected by the organization’s use of AI.

Effective third-party management extends beyond traditional vendor oversight. Organizations may need to demonstrate responsible AI practices through contractual controls, due diligence questionnaires, documentation reviews, and ongoing monitoring, as well as through client-facing disclosures and responses to inquiries.

These expectations often surface through contractual obligations, customer due diligence requests, and RFP processes – even in the absence of formal regulatory requirements. By integrating AI considerations into existing third-party and client engagement processes, organizations can provide greater transparency, manage risk more effectively, and build trust.

Looking Ahead

As AI technologies continue to evolve, expectations around their governance will continue to rise. While ISO/IEC 42001 is a voluntary standard rather than a regulation, it provides organizations with a structured and defensible approach to AI governance that aligns naturally with emerging legal, regulatory, and customer expectations.

Rather than prescribing specific controls or playbooks, ISO 42001 embeds AI governance into the same management system principles that have served as the bedrock of information security, privacy, and risk management programs for years. By applying these familiar principles to AI, organizations can establish governance practices that are consistent, scalable, and capable of adapting as AI use cases and risks change over time.

Visit our services page to learn how Resiliam can help with ISO 42001.

In October 2025, the International Standards Organization (“ISO”) released a second version of its flagship standard for privacy governance, ISO/IEC 27701.

The 2025 version, entitled “Information security, cybersecurity and privacy protection — Privacy information management systems — Requirements and guidance”, updates and replaces the first edition of the standard from 2019.

Similar to the original version, the 2025 standard provides a structured, certifiable framework for organizations to demonstrate accountability, manage risks, and continually improve their privacy practices around the processing of personally identifiable information (“PII”).

That said, the 2025 version differs from its predecessor in a number of material ways.

Independence from ISO/IEC 27001

The most substantive change in the 2025 version of ISO/IEC 27001 is that it is no longer an extension to ISO/IEC 27001, the leading international standard for information security. Under the 2025 version, organizations can establish and certify a standalone Privacy Information Management System (“PIMS”) without building upon an underlying Information Security Management System (“ISMS”). In other words, ISO/IEC 27701 certification no longer requires certification to ISO/IEC 27001:2022.

That said, given the significant overlap and alignment between the required information security controls and processes in ISO/IEC 27701 and those in ISO/IEC 27001, layering a PIMS on an ISMS is likely still the most effective approach for most organizations.

But by creating independence from ISO/IEC 27001, the International Standards Organization was able to restructure ISO/IEC 27701 with the same architecture as other ISO management system standards (i.e., identical clause numbers, similar clause descriptions, etc.), enabling seamless integration with a variety of other ISO standards, including those for business continuity and quality management, not just information security.

Modified Control Sets

Another key difference in the 2025 version is a reworking of the control sets included in the Annexes of the standard.

ISO/IEC 27701:2025 contains three sets of controls within its Annex A:

• 31 controls for PII Controllers (entities who determine the purposes and means of PII processing);
• 18 controls for PII Processors (entities who carry out PII processing on the instructions of a Controller); and
• 29 security controls for both PII Controllers and PII Processors.

Although the 2025 version has a new numbering scheme as compared to its predecessor, the content of the PII Controller and PII Processor controls in Annexes A.1 and A.2 were not materially changed.

The significant change comes in the form of Annex A.3, which now has mandatory information security control statements that are required for consideration of applicability, as opposed to implementation guidance only, which the 2019 version had contained.

These information security controls must be addressed by both PII Controls and PII Processors and are mapped to the 2022 version of ISO/IEC 27001.

Additional Changes

The 2025 version contains a number of other changes in relation to the prior release:

• The organization’s privacy policy must provide a framework for setting privacy objectives;
• The process for privacy risk treatment must identify and document elements of the organization’s information security program; and
• Implementation guidance for all three sets of controls in ISO/IEC 27701 is now consolidated into Annex B, which should be considered when determining the organization’s relevant PII Controller, PII Processor, and security controls.

Certification and Transition

Organizations seeking initial certification to ISO/IEC 27701 should absolutely implement the 2025 version of the standard rather than its predecessor. Furthermore, ISO has already released ISO/IEC 27706:2025, a standard for certification bodies that enables them to audit and certify to the 2025 version.

That said, the Global Accreditation Cooperation Incorporated (“Global ACI”) has not yet released guidance for organizations that were already certified to the 2019 version to transition to the 2025 version. Past experience indicates that a transition window of 18-36 months will likely be allowed. Resiliam will continue to monitor Global ACI releases for developments and will provide timely updates.