AI and Data Protection

AI and Data Protection

One AI system, two rulebooks: the AI Act and GDPR both apply.

What Is AI and Data Protection?

The EU AI Act and GDPR are independent but overlapping. When an AI system processes personal data, GDPR always applies, and so does the AI Act whenever the system falls within its scope. You work through a healthcare AI triage system: GDPR Article 22 rights on automated decisions, data minimization against vendor demands for more data, dual transparency duties and the difference between a DPIA and a FRIA, until a patient complaint tests whether your measures hold.

What You'll Learn in AI and Data Protection

AI and Data Protection — Training Steps

  1. Two Frameworks, One AI System

    The EU AI Act and GDPR overlap when AI processes personal data. GDPR covers any such system, and the AI Act applies too when the system is in its scope, as this triage system is. GDPR lawful basis - a valid legal ground for processing. GDPR Article 22 - the right not to be subject to purely automated decisions with significant effects. DPIA requirements - a Data Protection Impact Assessment may be required for high-risk processing. Data minimization - only data necessary for the purpose. The frameworks are enforced separately but create overlapping obligations.

  2. AI Triage System Overview

    Wellspring Health Services has developed an AI triage system to help manage patient appointments more efficiently. Alice signs in to the clinical portal and opens the system overview page to review how it works before conducting her compliance assessment.

  3. GDPR Article 22: Automated Decision-Making

    GDPR Article 22 gives individuals the right not to be subject to purely automated decisions with significant effects. EU AI Act Article 14 separately requires human oversight for high-risk AI. The two are complementary: GDPR creates a patient right, the EU AI Act creates a design obligation. Both must be satisfied independently.

  4. Lawful Basis for Health Data

    The triage system processes health data, which is a special category under GDPR Article 9. Standard processing grounds like legitimate interest are not sufficient for health data. The Article 9(2) conditions that fit this case are: Explicit consent (GDPR Art. 9(2)(a)) - the patient gives explicit, informed consent to the processing of their health data by the AI system. Necessary for medical treatment or diagnosis (GDPR Art. 9(2)(h)) - the processing is necessary for healthcare purposes and is carried out by a health professional or someone subject to professional secrecy obligations. Legitimate interest alone is never sufficient for processing health data under GDPR. Organizations must identify a valid Article 9(2) condition in addition to an Article 6 lawful basis.

  5. Mid-Exercise Check

  6. Data Minimization vs AI Performance

    The AI vendor has requested access to full patient medical records, arguing that broader data would improve triage accuracy. However, GDPR's data minimization principle (Article 5(1)(c)) requires that only data necessary for the specific purpose is processed. For symptom-based triage, the system needs current symptom descriptions and basic patient demographics. Full medical history, medication records, and past treatment history are excessive for determining appointment priority. Processing more data than necessary violates GDPR - even if it would improve AI performance.

  7. Patient Privacy Notice Review

    Alice opens the patient-facing privacy notice from the clinical portal. A single notice can satisfy both GDPR and EU AI Act, but only if it covers every required disclosure.

  8. DPIA vs FRIA

    This healthcare AI system may need two impact assessments: DPIA (Data Protection Impact Assessment) - under GDPR, when processing is likely to pose a high risk to individuals' rights. Covers data protection risks . FRIA (Fundamental Rights Impact Assessment) - under AI Act Article 27, before high-risk AI is used by public bodies, providers of public services such as healthcare, or deployers of credit-scoring or life and health insurance AI. Covers all fundamental rights : non-discrimination, dignity, access to healthcare, and an effective remedy. They can be done together, and the FRIA can reuse parts of the DPIA, but each must meet its own regulation's requirements.

  9. A Patient Complaint

    Dr. Sarah Park pings Alice on Telegram about a patient who is unhappy with her AI-assigned priority. Alice needs to advise Dr. Park on how to respond within the bounds of GDPR Article 22 and EU AI Act human-oversight obligations.

  10. Key Takeaways

    Here is what to remember when AI systems process personal data: Dual compliance often applies: AI systems that process personal data are always covered by GDPR, and by the EU AI Act as well when they fall within its scope, as this triage system does. These are independent frameworks with separate enforcement. Human oversight serves both frameworks: GDPR Article 22 gives people a right not to be subject to purely automated decisions, with safeguards such as human intervention (Article 22(3)), while EU AI Act Article 14 requires high-risk systems to be designed for effective human oversight. The legal reasons differ. They are complementary, not substitutes. Data minimization applies even when more data would help: Processing more data than necessary violates GDPR regardless of whether it improves AI performance. Privacy notices must satisfy both frameworks: GDPR requires transparency about data processing; the EU AI Act requires transparency about AI system usage. Both can be addressed in a single notice. Both DPIA and FRIA may be required: A DPIA focuses on data protection risks; a FRIA covers all fundamental rights. They serve different purposes even when conducted together. Patients have rights under both: GDPR gives access to information (Articles 13-15) and human intervention in purely automated decisions (Article 22(3)); the EU AI Act adds a right to an explanation of high-risk AI decisions (Article 86).

Security Framework Coverage

NIST CSF

  • PR.AT-01 Personnel are provided with awareness and training so that they possess the knowledge and skills to perform general tasks with cybersecurity risks in mind

NICE Framework

  • OG-WRL-008 Privacy Compliance
  • OG-WRL-002 Cybersecurity Policy and Planning

GDPR

  • Art. 22 Automated individual decision-making, including profiling
  • Art. 35 Data protection impact assessment
  • Art. 9 Processing of special categories of personal data

EU AI Act

  • Art. 26 Obligations of deployers of high-risk AI systems
  • Art. 27 Fundamental rights impact assessment for high-risk AI systems
  • Art. 86 Right to explanation of individual decision-making