European Data Protection Board (EDPB) Guidelines 3/2019 - definition
European Data Protection Board (EDPB) Guidelines 3/2019 are a European Union interpretive document on the processing of personal data through video devices. Their full title is Guidelines 3/2019 on processing of personal data through video devices. The EDPB adopted the first version on 10 July 2019 and Version 2.0 on 29 January 2020 after public consultation.
The Guidelines explain how Regulation (EU) 2016/679, the General Data Protection Regulation (GDPR), applies to video surveillance, recorded footage, cameras, dashboard cameras, doorbell cameras, and similar image-based systems. They are not a regulation, but they are a key supervisory authority interpretation for organizations that capture, review, disclose, anonymize, or publish images and video where the GDPR applies.
For image and video anonymization, the central point is simple. If a person can be identified from a face, clothing, location, behavior, vehicle license plate, or other visual context, the recording can contain personal data under GDPR Article 4(1). Blurring or masking faces and license plates is therefore a common control used to reduce identifiability before footage is shared, archived, used for training, or published.
Legal scope of EDPB Guidelines 3/2019
The Guidelines apply where video processing falls within the material and territorial scope of the GDPR. They should be read together with GDPR Article 2 on material scope, Article 3 on territorial scope, Article 4 on definitions, and Recital 26 on anonymous information.
Issue | EU legal reference | Practical meaning for video anonymization |
|---|---|---|
Personal data | GDPR Article 4(1), Recital 26 | Faces, license plates, and contextual visual identifiers can make footage personal data if a person is identifiable. |
Processing | GDPR Article 4(2) | Recording, storing, viewing, exporting, blurring, and deleting video are processing operations. |
Household exemption | GDPR Article 2(2)(c), Recital 18 | Purely personal or household use may be outside the GDPR, but the exemption is narrow. Public-facing or work-related recording usually needs assessment. |
Anonymous information | GDPR Recital 26 | Footage is outside the GDPR only if individuals are no longer identifiable by means reasonably likely to be used. |
Biometric data | GDPR Article 4(14), Article 9, Recital 51 | Video is not automatically biometric data. It becomes biometric data when processed by specific technical means for unique identification or authentication. |
Key exceptions and limits
EDPB Guidelines 3/2019 clarify that not every video recording is subject to the same compliance burden. The scope depends on the purpose, the setting, the identifiability of people in the footage, and whether the GDPR applies at all.
- Personal or household activity. A private individual recording only for personal household purposes may fall outside the GDPR under Article 2(2)(c). This exception becomes difficult to rely on when a camera captures public space, employees, customers, neighbors, or visitors.
- Law enforcement processing. Processing by competent authorities for law enforcement purposes is generally governed by Directive (EU) 2016/680 rather than the GDPR.
- Truly anonymous footage. If faces, license plates, and other identifiers are irreversibly removed or masked so that identification is no longer reasonably likely, the result may fall outside the GDPR under Recital 26.
- Special categories of data. Ordinary video is not automatically special category data. However, video analytics used to uniquely identify a person can trigger GDPR Article 9 rules on biometric data.
Role of EDPB Guidelines 3/2019 in face and license plate blurring
The Guidelines support a risk-based approach to reducing visual identifiability. In practice, this means that organizations should decide whether they need identifiable footage, or whether blurred footage is sufficient for the purpose.
Face blurring is relevant because a visible face is often enough to identify a person directly or indirectly. License plate blurring is also relevant in the European Union because vehicle registration numbers can identify or help identify a driver, owner, employee, visitor, or customer. In many European deployments, license plate masking is commonly used or expected in practice, depending on the combined effect of GDPR principles, national privacy rules, civil law protections, and supervisory authority expectations.
For EU publications and external sharing, face anonymization must also be assessed against national image rights and civil law rules. Potential exceptions or permissions may apply where the person is a public figure, the person is only incidental to a wider public scene such as a concert, sports event, or assembly, or the person has consented or contracted for use of their image. These exceptions do not remove the need to assess GDPR requirements when personal data is processed.
Compliance requirements for video processing
EDPB Guidelines 3/2019 connect video devices to the core GDPR obligations. The following controls are especially important before exporting or disclosing footage to another entity.
- Purpose limitation. Define why footage is processed before collection or anonymization, as required by GDPR Article 5(1)(b).
- Data minimization. Capture and retain only what is necessary under GDPR Article 5(1)(c). Blurring faces and license plates can support this principle.
- Lawful basis. Identify a lawful basis under GDPR Article 6, often legitimate interests under Article 6(1)(f), consent under Article 6(1)(a), or legal obligation under Article 6(1)(c), depending on the case.
- Transparency. Provide layered notices under GDPR Articles 12, 13, and 14 where required.
- Storage limitation. Retain identifiable footage only as long as necessary under GDPR Article 5(1)(e). The EDPB states in Guidelines 3/2019 that storage beyond a short period needs stronger justification.
- Security. Protect footage against unauthorized access, alteration, and disclosure under GDPR Article 32.
- Data protection by design. Build masking, access controls, and deletion into the workflow under GDPR Article 25.
- Data protection impact assessment. A Data Protection Impact Assessment (DPIA) may be required under GDPR Article 35, especially for systematic monitoring of publicly accessible areas on a large scale.
Technical application in anonymization workflows
In an anonymization workflow, the Guidelines are relevant before and after technical processing. They affect what footage may be collected, who may review it, what must be blurred, how long the original may be kept, and whether the output can still identify a person.
Workflow stage | Compliance question | Technical control |
|---|---|---|
Import | Is there a lawful basis and defined purpose? | Limit access to authorized staff and process only required files. |
Detection | Which visual identifiers must be removed? | Use face detection and license plate detection for automatic masking. |
Review | Are any identifiers still visible? | Use manual review for missed faces, plates, logos, tattoos, name badges, documents, or monitor content. |
Export | Can the output still identify a person? | Export a masked copy and control retention of the original. |
Gallio PRO uses face detection, not facial recognition. Automatic blurring covers faces and license plates only. It does not automatically detect company logos, tattoos, name badges, documents, or content displayed on monitors, but these elements can be blurred manually in the built-in editor. Gallio PRO is not a real-time anonymization or video stream anonymization system, and it does not collect logs containing face detections, license plate detections, personal data, or sensitive data.
Key metrics and parameters
EDPB Guidelines 3/2019 do not define computer vision accuracy thresholds. For compliance, however, technical performance should be measured and documented because missed detections can leave personal data visible.
Parameter | Meaning | Why it matters |
|---|---|---|
Recall | Detected relevant objects divided by all relevant objects in the test set | Low recall means faces or license plates may remain visible. |
Precision | Correct detections divided by all detections made by the system | Low precision increases unnecessary masking and manual review time. |
False negative rate | Missed relevant objects divided by all relevant objects | This is the most important risk metric for anonymization quality. |
Retention period | Time for which identifiable source footage is kept | Longer retention requires stronger necessity and security justification under GDPR Article 5(1)(e). |
US equivalent
There is no direct United States equivalent to EDPB Guidelines 3/2019. The closest frameworks are sectoral and state-specific, including the California Consumer Privacy Act (CCPA) as amended by the California Privacy Rights Act (CPRA), the Illinois Biometric Information Privacy Act (BIPA), 740 ILCS 14, the Health Insurance Portability and Accountability Act (HIPAA) for covered health contexts, and the Family Educational Rights and Privacy Act (FERPA) for education records. The United States has no single federal rule that mirrors the GDPR approach to video devices. A US organization receiving footage from an EU entity should treat the EDPB Guidelines as an EU compliance reference when the GDPR applies.
Standards and references for EDPB Guidelines 3/2019
The most reliable sources are the EDPB text and the GDPR itself. Technical teams should also align anonymization testing with documented quality metrics and information security controls.
- European Data Protection Board, Guidelines 3/2019 on processing of personal data through video devices, Version 2.0, adopted 29 January 2020.
- Regulation (EU) 2016/679, General Data Protection Regulation, especially Articles 2, 3, 4, 5, 6, 9, 12-15, 17, 21, 25, 32, and 35, and Recitals 18, 26, 47, 51, and 91.
- Directive (EU) 2016/680 on processing of personal data by competent authorities for law enforcement purposes.
- International Organization for Standardization and International Electrotechnical Commission, ISO/IEC 27001:2022, information security management systems.