Skip to content
All articles
Law & regulation 7 min read· by Lars Zimmermann

AI officer and data protection officer: one person for both?

Can the data protection officer also be the AI officer? Where the roles differ, when a conflict of interests arises and how to separate them cleanly.

In short

The data protection officer is a statutory role with freedom from instructions and protection against dismissal. They monitor and advise. The AI officer is a voluntary role that shapes and approves. One person can be both as long as they do not decide, as AI officer, on the purposes and means of processing personal data. Otherwise a conflict of interests under Article 38(6) GDPR looms.

Auf Deutsch lesen: deutsche Fassung

In many companies the obvious solution is already on the table. AI has to do with data. For data we have the data protection officer. So they take on AI as well. That saves a role, a round of coordination and an invoice.

It can work. It can also damage exactly the independence for which the data protection officer is in the law. Where the line runs was described by the Court of Justice of the European Union in 2023. First, a look at how differently the two roles are built.

Two roles, two designs

The data protection officer is a role of the law. The GDPR sets out in three articles when one must be designated (Article 37), what position they hold (Article 38) and which tasks they have at least (Article 39). In Germany, Section 38 of the Federal Data Protection Act adds that designation is mandatory, as a rule, from 20 people constantly engaged in the automated processing of personal data.

The AI officer is a role of the company. No law prescribes it and no law describes it. The EU AI Act does not mention it. What the person does and may do is written in the mandate management gives them. If nothing is written there, they may do nothing.

  • Instructions. The data protection officer receives no instructions regarding the exercise of their tasks (Article 38(3) GDPR). The AI officer is bound by instructions like any other function unless the mandate says otherwise.
  • Protection. The data protection officer may not be dismissed or penalised for performing their tasks. Where designation is mandatory in Germany, Sections 38(2) and 6(4) of the Federal Data Protection Act add special protection against removal and dismissal. The AI officer has no such protection.
  • Reporting line. The data protection officer reports directly to the highest management level. For the AI officer the mandate must set this.
  • Subject matter. The data protection officer looks at personal data. The AI officer looks at all AI tools, including those that only process machine data or drawings.

The last point is often overlooked. A tool that suggests cutting parameters from machine logs has little to do with data protection. It has a lot to do with AI risk: wrong suggestions, trade secrets with the provider, nobody checking the output. For the data protection officer this tool is outside their remit.

A practical rule follows for the dual role. The statutory protection attaches to the task as data protection officer. It does not extend to the AI task. You can reassign AI coordination if it is better placed elsewhere. One limit remains: the reassignment must not be a response to uncomfortable data protection work. Otherwise it touches the prohibition on penalising the officer in Article 38(3) GDPR.

One monitors, the other shapes

The most important difference lies in the kind of work. Article 39 GDPR gives the data protection officer tasks that all point in one direction: inform, advise, monitor compliance, advise on the data protection impact assessment on request, cooperate with the supervisory authority. They decide nothing. They say whether a decision holds.

An AI officer works differently. They keep the inventory of tools. They write the policy. They approve new tools or reject them. They help determine which data may go into which tool. That is shaping. Depending on the mandate it is also deciding.

As long as the two roles sit with different people, this complements well. The AI officer proposes approving a tool for applicant correspondence. The data protection officer checks whether processing the applicant data in this way is lawful. Two pairs of eyes, two angles.

Where the conflict of interests begins

The GDPR expressly allows the data protection officer other tasks. Article 38(6) sets one condition: those tasks must not result in a conflict of interests. The company has to ensure that.

What a conflict of interests is was described by the Court of Justice on 9 February 2023 (Case C-453/21). A conflict may exist where a data protection officer is entrusted with other tasks or duties which would result in them determining the objectives and methods of processing personal data. Whether that is the case must be assessed case by case in the light of all circumstances, in particular the organisational structure and internal rules.

Applied to the dual role: what counts is the mandate the person holds as AI officer. The title decides nothing. Two cases show the difference.

  • Uncritical: the person keeps the inventory, organises briefings, collects questions and prepares approvals. Management or the department decides on the approval. The person sets no purposes and means. They coordinate.
  • Critical: the person themselves approves which AI tool may process customer data or applicant data. They help select the provider and determine which data goes in. Afterwards they are supposed to monitor, as data protection officer, whether this processing is lawful. They are checking their own decision.

The second case is the one the Court describes. Everyone knows the pattern from auditing: nobody audits their own work. For the data protection officer this principle is in the law.

The company bears the consequences. A breach of Article 38 is among the controller obligations for which Article 83(4) GDPR provides fines. Whether a conflict exists in a specific case is a legal question. In case of doubt it belongs with a lawyer or the supervisory authority. I give a professional assessment here, from an auditor’s point of view. This is not legal advice.

Three models that hold

  • Separate people. The AI officer shapes and approves, the data protection officer monitors. This is the clearest solution and the right one for companies with many tools or their own AI development.
  • One person, limited mandate. The data protection officer takes on AI coordination: inventory, briefings, preparation. Approvals and the decision on data and providers lie in writing with management or the department. This fits small companies with a few standard tools.
  • Committee. A small group from IT, the business, the data protection officer and management decides on approvals. The data protection officer sits at the table in an advisory capacity and does not vote.

For the second solution the limit must be in the mandate, in sentences nobody has to interpret. For example:

  • “The AI officer keeps the AI inventory and prepares approvals.”
  • “Management decides on the approval of an AI tool.”
  • “The AI officer does not decide on the purposes and means of processing personal data.”
  • “In case of doubt, the tasks as data protection officer take precedence.”

The second solution also has a practical problem: time. An external data protection officer has a fixed budget, an internal one usually has a second main job. AI coordination comes on top. Write into the mandate how many hours are planned for it. Otherwise both suffer in the end.

If your external data protection officer offers the AI role as a package, three questions are worth asking. Who decides on approvals in this package? What experience is there with AI tools without personal data, for example in design or manufacturing? And what happens to the AI role if the data protection contract ends?

Where the two must work together

Separate roles do not mean separate work. The two need each other in five places:

  • Records. The AI inventory and the record of processing activities under Article 30 GDPR overlap for every tool with personal data. A shared field in the AI inventory that points to the entry in the record of processing is enough.
  • Approvals. Every approval request for a tool with personal data goes to the data protection officer before the decision.
  • Impact assessment. For processing likely to result in a high risk, Article 35 GDPR requires a data protection impact assessment. Anyone working to ISO/IEC 42001 also carries out an AI system impact assessment (Annex A, A.5.2). Both ask similar questions of the same tool. Do them in one session.
  • Providers. The data processing agreement under Article 28 GDPR and the AI officer’s questions to the provider belong in one folder.
  • Training. The briefing on AI tools and the data protection training share a core: which data may go where? A joint session saves time and prevents two people giving two different answers.

Companies below the threshold in Section 38 of the German Federal Data Protection Act often have no data protection officer at all. The question of the dual role does not arise then. The data protection questions remain, because the GDPR applies regardless of designation. In that case the AI officer needs someone to ask about data protection, in-house or external.

One person can wear both hats as long as they do not decide under one what they are supposed to check under the other.

What you can do on Monday: if the AI role sits with your data protection officer, read what they may decide as AI officer. If it includes approval of tools with personal data, move that decision to management. It costs one signature and takes the biggest risk out of the dual role.

What the AI role covers in practice and how it fits into AI governance:

The AI officer: does your company need the role?
Share: LinkedIn E-Mail

Frequently asked questions

Can the data protection officer also be the AI officer?+

Yes, the GDPR allows the data protection officer other tasks as long as no conflict of interests arises (Article 38(6)). It becomes critical when, as AI officer, they themselves decide on the purposes and means of processing personal data, for example by approving a tool.

What is the difference between an AI officer and a data protection officer?+

The data protection officer is regulated by law, free from instructions and protected against dismissal for performing their tasks. They monitor and advise on personal data. The AI officer is a voluntary role without statutory protection. They shape how all AI tools are used, including those without personal data.

What did the CJEU decide on conflicts of interests?+

In its judgment of 9 February 2023 (C-453/21) the Court held that a conflict of interests may exist where a data protection officer is entrusted with tasks which would result in them determining the objectives and methods of processing. The assessment is made case by case.

Must the data protection officer be involved with AI tools?+

Yes, as soon as a tool processes personal data. Under Article 38(1) GDPR they must be involved, properly and in a timely manner, in all issues relating to the protection of personal data. That applies regardless of who is the AI officer.

Author & expert review: Lars Zimmermann · ISO/IEC 42001 Senior Lead Auditor & Senior Lead Implementer · ISO/IEC 27001 Lead Auditor & Lead Implementer (PECB)

Last updated: 10 November 2026. Researched and reviewed to the best of our knowledge; not a substitute for individual legal advice.

Questions about your own case?

In a free 15-minute intro call we assess where you stand on ISO 42001, ISO 27001 and the EU AI Act, honestly and without a sales pitch.

Continue reading