The Skill Explained
What Are Privacy and GDPR?
Privacy and GDPR covers protecting personal data and following the laws about it. The GDPR (General Data Protection Regulation) is the European Union's data protection law; HIPAA protects health information in the United States.
Why AI Employers Ask for Them
AI systems often process personal data: customer messages, patient records, employee files. Getting it wrong brings fines, lost trust and blocked projects. Employers want AI builders who design privacy in from the start and can work with legal and data protection teams.
How It Works
The GDPR sets principles: process personal data lawfully and for a stated purpose, collect only what is needed (data minimisation), keep it accurate and no longer than necessary, and secure it. People have rights to access, correct and delete their data. Health data is a special category with stricter rules, and a Data Protection Impact Assessment is needed for high-risk processing.
Privacy by design turns principles into engineering: remove or mask personal details before data reaches a model, pseudonymise identifiers, control access, encrypt data, keep audit logs and set retention limits. Contracts with AI providers must cover how data is used, for example that it is not used to train their models.
Privacy-enhancing technologies go further: differential privacy adds calibrated noise to protect individuals in statistics, and federated learning trains models without moving raw data off its source.
What You Do With It at Work
- Map what personal data an AI feature uses and why.
- Remove or mask personal data before it reaches a model.
- Set access controls, encryption and retention rules.
- Support Data Protection Impact Assessments with technical facts.
- Check AI vendors' data terms with the privacy team.
Tools and Terms You Will Meet
- Personal data
- Any information about an identifiable person.
- Data minimisation
- Using only the data needed for the purpose.
- Pseudonymisation
- Replacing identifiers so data cannot be linked to a person without extra information.
- DPIA
- Data Protection Impact Assessment, required for high-risk processing.
- HIPAA
- The US law protecting health information.
Industry Example: Pharma
Before free-text adverse event reports are used to fine-tune a model, names, dates of birth and contact details are automatically masked, and the processing is documented in a DPIA approved by the data protection officer.
A Common Mix-Up
Anonymised and pseudonymised data are not the same. Pseudonymised data can still be linked back to people and stays personal data under the GDPR; truly anonymised data cannot.
How Deep Do You Need to Go?
- Know itExplain personal data and the main GDPR principles.
- Use itBuild privacy steps such as masking, access control and retention into AI systems.
- Master itLead privacy by design and assessments for AI programmes.