ISO 42001 vs ISO 27701: AI management and privacy management

ISO 42001 and ISO 27701 may seem far apart at first glance. One covers AI, the other privacy. In practice, however, they overlap. AI systems almost always process personal data: training data, user data and inference outputs. That is exactly where the two standards meet. On this page we explain the overlap, describe the differences and answer when you need both.

For a wider comparison of ISO 42001 with information security standards, see ISO 42001 vs ISO 27001. ISO 27701 is a privacy management extension to ISO 27001. For an introduction to ISO 27701, see the ISO 27701 overview page.

What is ISO 27701?

ISO 27701 is a privacy management extension to ISO 27001. It extends the information security management system (ISMS) with privacy-specific requirements and guidance. Its goal is to show that you comply with privacy legislation such as the Dutch AVG (General Data Protection Regulation, GDPR).

ISO 27701 introduces two new roles: the Privacy Information Management System (PIMS) controller, for organisations that decide why personal data is used, and the PIMS processor, for organisations that process personal data on behalf of others.

ISO 27701 is not a standalone certifiable standard. You implement it as an extension of an existing ISO 27001 system.

What are the similarities?

Both standards are built on the same High Level Structure , which simplifies integration. The similarities go deeper than structure, though.

Personal data as a shared theme. ISO 42001 requires attention to privacy as an ethical principle and as a risk factor. ISO 27701 goes further into how you protect that privacy in day-to-day operations. If your AI systems process personal data, both standards have something to say.

Data governance. Both standards set requirements for how you handle data: origin, quality, retention and access. The starting point differs (privacy versus AI risk), but the measures overlap strongly.

Transparency towards data subjects. ISO 42001 asks for explainability of AI decisions. ISO 27701 asks for transparency about data processing. For AI systems that make decisions based on personal data, both apply.

Rights of data subjects. ISO 27701 turns the AVG rights (access, correction, objection) into processes. ISO 42001 asks for mechanisms so people can challenge AI decisions. With automated decision-making, these requirements meet.

Theme	ISO 27701	ISO 42001
Data governance	Extensive, focused on the AVG	AI-specific (training data, data quality)
Transparency	About data processing	About how AI works and decides
Rights of data subjects	Access, correction, objection (AVG)	Explainability, challengeability of AI decisions
Risk analysis	Privacy Impact Assessment (PIA/DPIA)	AI System Impact Assessment (AISIA)
Governance	Privacy officer / FG (functionaris gegevensbescherming, DPO)	AI officer / AI governance
Data retention	Retention policy	Lifecycle of training data

What are the differences?

ISO 27701 is a tool for AVG compliance. It is designed explicitly to help show compliance with the AVG. It contains a detailed mapping to AVG articles, rights of data subjects and processor obligations. ISO 42001 has no explicit AVG mapping.

ISO 42001 covers AI-specific risks. Model drift, bias, unexplainable decisions and misuse of AI are AI risks that do not belong in a privacy management system. ISO 27701 does not address them.

Impact analyses differ in nature. A DPIA (Data Protection Impact Assessment) under ISO 27701 assesses the privacy risks of a processing activity. An AI impact assessment under ISO 42001 assesses the social impact of an AI system on individuals and groups. They complement each other, but they are not interchangeable.

ISO 27701 requires an ISMS as a foundation. You cannot implement ISO 27701 on its own. ISO 42001 can be set up as a standalone certification, even without ISO 27001 or ISO 27701.

ISO 42001 Annex D explicitly refers to ISO 27701 as relevant for organisations whose AI systems process personal data. The standard therefore recognises that the two standards are used as a complementary set.

When do you need both?

Only ISO 42001

You use AI but barely process personal data. Examples include AI for process optimisation without personal data input, or industrial AI such as predictive maintenance on machines.

In practice this is rare: almost every AI application processes personal data at some point.

Only ISO 27701

You process a lot of personal data but use little or no AI. An example is an organisation that runs an extensive CRM system without any AI functionality.

Both standards

You process personal data through AI systems. This is the most common scenario.

Concrete cases where you need both:

HR technology company that uses AI for recruitment and selection (personal data plus high-risk AI)
Fintech with AI-driven credit assessment (financial personal data plus high-risk AI)
Healthcare organisation with AI-supported diagnostics (special category personal data plus high-risk AI)
Marketing platform with personalised recommendations (behavioural data plus profiling)
SaaS company with AI features in its product (customer data plus processor role)

The AVG has a specific right to an explanation for fully automated decisions (Article 22 AVG). If your AI makes automated decisions with legal effects for people, both AVG compliance (ISO 27701) and explainability of AI (ISO 42001) are required.

How do you combine ISO 42001 and ISO 27701?

Your starting point determines the strategy.

You already have ISO 27001 and want to add ISO 27701 and ISO 42001

This is the most efficient route for organisations that take AI governance and privacy seriously.

Build it in two steps:

Add ISO 27701 as an extension to your ISMS (relatively quick, as the structure is already in place)
Add ISO 42001 as a complementary AIMS, with shared governance (policy, roles, audit)

Where processes overlap, integrate them. The DPIA and the AI impact assessment can be combined into one process for AI systems that process personal data.

You have nothing yet and want to start

Start with ISO 27001 as the foundation. Add ISO 27701 when you want to formalise AVG compliance, and build ISO 42001 on top when AI is the core of your activities.

Or build an integrated system that covers all three straight away. This requires more planning, but it prevents tripling your work.

The shared elements

Things you only need to set up once:

Governance structure (policy, roles, management commitment)
Risk management process (methodology, risk appetite)
Internal audit programme
Management review
Document control
Supplier management

Things you need to do twice (but can integrate):

Risk register: both privacy risks and AI risks
Impact analyses: DPIA and AISIA (can be combined for AI systems that process personal data)
Controls: Annex A of ISO 27001/27701 and Annex A of ISO 42001

The practical link: AI systems and personal data

The most direct overlap lies in the processing of personal data by AI systems. A number of concrete situations:

Training data containing personal data. If you train a model on customer data, employee data or other personal data, the AVG applies to that processing. ISO 27701 helps you ensure lawful processing, purpose limitation and minimal data retention. ISO 42001 asks for data quality, representativeness and bias assessment of that same data.

Inference on personal data. If your AI system makes decisions based on personal data (credit, recruitment, diagnosis), the explainability requirement of ISO 42001 and the transparency obligation of ISO 27701 / AVG both apply.

Profiling. Profiling people based on behaviour is an AVG-sensitive processing activity (Article 4(4) and Article 22 AVG). ISO 27701 requires a legal basis and transparency. ISO 42001 requires an assessment of bias and impact.