On April 20, 2026, the Spanish Data Protection Agency (AEPD) has published new guidance on how to comply with the GDPR when using AI‑powered voice transcription tools. The guidance builds on earlier AEPD guidance on this topic from January 2026. This blog post sets out the key takeaways of both guidance documents, which are only available in Spanish.

The AEPD’s guidance confirms a risk‑based approach to AI‑powered voice transcription. Organizations using these tools should not treat transcription as a purely technical feature, but as a processing activity that requires continuous governance, clear transparency, and proactive safeguards. Given the widespread and growing use of transcription tools across business functions, this guidance is likely to be relevant well beyond Spain.

Key Takeaways

  • A person’s voice will generally constitute personal data under the GDPR, except where it is fully anonymized or purely synthetic; however, the guidance does not specify how voice anonymization should be achieved in practice. In any event, related metadata and the content of the communication may still constitute personal data.
  • Organizations using AI transcription tools act as controllers. This applies regardless of whether the tool is developed in‑house or provided by a third‑party vendor, as the organization determines the purposes and means of the processing, i.e., transcription.
  • Controllers must perform data protection diligence when selecting transcription tools and vendors. Controllers should select only providers that offer clear information on any additional processing beyond transcription, such as reuse of voice data or transcripts to train or improve AI models, human review for quality assurance or model tuning, or other secondary uses, as well as on confidentiality and security measures, data retention and minimization, and data location. This due diligence is necessary to comply with the GDPR in general, including the conditions under which a controller may engage a data processor under Article 28 GDPR. Automation does not reduce GDPR obligations, and the use of AI‑based transcription tools increases expectations around accountability, transparency, and data subject rights where such risks are foreseeable.
  • Due diligence is required throughout the lifecycle of the tool. The guidance states that this obligation applies to both controllers and processors. They must assess risks on a continuous basis, including risks linked to transcription errors, linguistic bias, and the inference of sensitive information. Passive reliance on AI functionality is insufficient.
  • Controllers must assess whether transcription tools makes inference based on voice data. Organizations must determine whether the tool infers emotions, thoughts, beliefs, health status, biometric identifiers, or other sensitive characteristics. Transcription services that include emotion detection or infer special categories of personal data are subject to strict GDPR requirements and may, in some cases, be prohibited under the EU AI Act (Regulation (EU) 2024/1689).
  • AI model retraining may shift roles and legal bases. Where voice data or transcripts are reused to retrain or improve an AI transcription system, the entity performing that retraining typically acts as a separate data controller and must establish a legal basis for doing so. In practice, this may involve supervised training, including human review and manual transcription by third parties, which must be sufficiently disclosed.
  • Users must be clearly informed before voice transcription starts and is occurring. Organizations must explain that transcription will take place, whether the data will be used for anything else, whether third parties may listen to the recordings (for example to improve the tool), what risks are involved, and how individuals can correct or delete their data. Transparency must also be continuous during the recording. Informing participants only before a session begins is not enough. Data subjects must remain aware that recording and transcription are taking place, for example through visible or audible indicators throughout the session.
  • Transcripts are not neutral texts but representations attributed to specific individuals. Automated transcripts form part of a processing activity in which controllers must comply with all GDPR principles in Article 5, including accuracy. Where a transcription wrongly attributes statements or details to a person, this is not just a technical error, and the controller must correct it without undue delay, as required by Article 16 GDPR.
  • Errors must be anticipated, not just corrected. Because the limitations of AI transcription systems are known, controllers must act proactively and not wait for errors to occur. This includes putting measures in place to prevent, detect, and correct inaccuracies, such as explaining the system’s limits to users, ensuring human review where appropriate, having clear correction procedures, and making it easy for individuals to exercise their rights of access and rectification.
  • The right of access applies even in multi‑speaker recordings. Individuals must be given effective access to their personal data under Article 15 GDPR, even where transcripts or recordings also contain third‑party data that is necessary to understand the context of what was said. Access cannot be refused solely for this reason. Where necessary, controllers should rely on technical measures such as anonymization, partial masking, or similar techniques to protect others’ rights. The AEPD notes that this approach is consistent with AEPD case law on audiovisual data and with the EU Data Act’s right of access to data generated by connected devices.
  • Consent must be specific to each recording session. The AEPD notes that that a general notice stating that joining a session implies consent is insufficient to establish valid consent. Consent cannot be generic or open‑ended. It must expire at the end of the relevant session, and controllers must ensure that recording automatically stops once the legal basis no longer applies.

*              *              *

The Covington team is closely monitoring regulatory developments related to AI‑powered voice transcription and broader data protection requirements. Similar issues are beginning to attract attention in other Member States, including recent guidance issued by the Swedish Supervisory Authority (IMY) on the use of AI‑based voice transcription tools in the public sector (only available in Swedish). Please reach out to a member of the team if you would like to discuss the AEPD’s guidance, assess its impact on your use of transcription tools, or need assistance with compliance.

Photo of Dan Cooper Dan Cooper

Daniel Cooper is co-chair of Covington’s Data Privacy and Cyber Security Practice, and advises clients on information technology regulatory and policy issues, particularly data protection, consumer protection, AI, and data security matters. He has over 20 years of experience in the field, representing…

Daniel Cooper is co-chair of Covington’s Data Privacy and Cyber Security Practice, and advises clients on information technology regulatory and policy issues, particularly data protection, consumer protection, AI, and data security matters. He has over 20 years of experience in the field, representing clients in regulatory proceedings before privacy authorities in Europe and counseling them on their global compliance and government affairs strategies. Dan regularly lectures on the topic, and was instrumental in drafting the privacy standards applied in professional sport.

According to Chambers UK, his “level of expertise is second to none, but it’s also equally paired with a keen understanding of our business and direction.” It was noted that “he is very good at calibrating and helping to gauge risk.”

Dan is qualified to practice law in the United States, the United Kingdom, Ireland and Belgium. He has also been appointed to the advisory and expert boards of privacy NGOs and agencies, such as the IAPP’s European Advisory Board, Privacy International and the European security agency, ENISA.

Read more about Dan Cooper
Show more Show less
Photo of Kristof Van Quathem Kristof Van Quathem

Kristof Van Quathem advises clients on information technology matters and policy, with a focus on data protection, cybercrime and various EU data-related initiatives, such as the Data Act, the AI Act and EHDS.

Kristof has been specializing in this area for over twenty…

Kristof Van Quathem advises clients on information technology matters and policy, with a focus on data protection, cybercrime and various EU data-related initiatives, such as the Data Act, the AI Act and EHDS.

Kristof has been specializing in this area for over twenty years and developed particular experience in the life science and information technology sectors. He counsels clients on government affairs strategies concerning EU lawmaking and their compliance with applicable regulatory frameworks, and has represented clients in non-contentious and contentious matters before data protection authorities, national courts and the Court of the Justice of the EU.

Kristof is admitted to practice in Belgium.

Read more about Kristof Van Quathem
Show more Show less
Photo of Anna Sophia Oberschelp de Meneses Anna Sophia Oberschelp de Meneses

Anna Sophia Oberschelp de Meneses advises on EU data protection, cybersecurity, and consumer law. Her practice covers the full range of Europe’s digital regulatory framework, including GDPR, ePrivacy, NIS2, the Cyber Resilience Act, the AI Act, the Digital Services Act, the Data Act…

Anna Sophia Oberschelp de Meneses advises on EU data protection, cybersecurity, and consumer law. Her practice covers the full range of Europe’s digital regulatory framework, including GDPR, ePrivacy, NIS2, the Cyber Resilience Act, the AI Act, the Digital Services Act, the Data Act, the European Health Data Space, and EU consumer protection law, including product safety, product liability, and consumer rights legislation. She focuses on the operational side of compliance — helping clients design policies and processes, draft documentation, and build the internal frameworks needed to meet regulatory requirements in practice.

She also advises on contentious matters, drawing on experience managing investigations before national regulators and proceedings before national courts and the Court of Justice of the European Union. She works closely with Covington’s disputes teams on matters at the intersection of regulatory compliance and litigation.

Read more about Anna Sophia Oberschelp de MenesesAnna Sophia's Linkedin Profile
Show more Show less