What is the Accountability Principle?

Accountability principle - definition

The accountability principle is a requirement of the European Union General Data Protection Regulation (GDPR). It requires a controller to comply with the data protection principles and to be able to demonstrate that compliance. Article 5(2) GDPR states: The controller shall be responsible for, and be able to demonstrate compliance with, paragraph 1 (‘accountability’).

For image and video anonymization, accountability means more than applying a blur effect to faces or license plates. The organization must be able to show why the footage was processed, which safeguards were selected, how the anonymization workflow was tested, who approved it, and whether the resulting files can still identify individuals. This evidence should be available to a supervisory authority, customer, auditor, or affected individual where relevant.

The principle applies to the controller, meaning the organization that determines the purposes and means of processing personal data. A processor that anonymizes footage on the controller’s behalf must support the controller through contractual and technical measures required under Article 28 GDPR.

Why accountability matters for face and license plate blurring

Images and video footage can contain personal data when people are identifiable directly or indirectly. A visible face, a vehicle license plate, the location of recording, and associated metadata may enable identification in context. Footage can also involve biometric data under Article 4(14) GDPR when it is processed through specific technical means for the purpose of uniquely identifying a natural person.

Face detection and license plate detection are commonly used to locate areas that require masking. However, detection performance alone does not prove anonymization. The controller should assess whether the final output prevents identification by reasonably likely means, taking account of the context, the recipients, and other available information.

Recital 26 GDPR explains that data protection principles do not apply to anonymous information. The same recital requires an assessment of means reasonably likely to be used for identification. Therefore, the original unblurred footage remains personal data while it is retained or accessible. A blurred export may be anonymous only if re-identification is no longer reasonably likely in its intended context.

Documenting an image and video anonymization process

A defensible record should describe the complete workflow, not only the final file. Documentation must connect the processing purpose, the source footage, the technical configuration, quality assurance results, and retention decisions.

  • Processing purpose: State why the footage is collected, anonymized, shared, or published. Examples include training materials, incident review, public reporting, or dataset preparation.
  • Footage scope: Record the source, recording date, locations, camera types, duration, resolution, frame rate, and expected categories of visible personal data.
  • Risk assessment: Identify risks such as missed faces, missed license plates, profile views, occlusion, motion blur, reflections, low light, and identification from non-blurred contextual details.
  • Technical method: Record the software version, detection settings, blur or masking method, tracking behavior, manual edits, and export settings.
  • Validation evidence: Keep the sampling method, reviewer instructions, review results, defect records, corrective actions, and approval decision.
  • Retention and access: Define who can access original footage, how long it is retained, where it is stored, and when it is deleted or securely archived.

Gallio PRO can support a documented workflow by detecting faces and license plates for automatic blurring and by enabling manual editing of areas that require additional masking. It does not automatically detect logos, tattoos, name badges, documents, or content displayed on monitors. These elements can be blurred manually in the built-in editor where necessary. Gallio PRO does not perform real-time or video-stream anonymization and does not collect logs containing face or license plate detections.

Validation metrics and evidence

The GDPR does not prescribe a minimum detection rate, blur strength, or sample size for anonymized video. Controllers must instead select controls that are appropriate to the risk, as required by Articles 24, 25, and 32 GDPR. Validation criteria should be established before release and applied consistently.

Measure

Formula or method

Accountability value

Detection recall

Correctly masked faces or plates / all faces or plates requiring masking

Shows the rate of missed items in a reviewed sample.

Detection precision

Correctly masked faces or plates / all automatically masked areas

Shows unnecessary masking and supports configuration review.

Manual correction rate

Frames or objects corrected manually / reviewed frames or objects

Identifies footage conditions that reduce automated performance.

Review coverage

Reviewed duration or frames / total duration or frames

Documents the extent of quality assurance.

Records should distinguish automated detection results from human review results. A missed face or license plate should trigger a documented correction and, where appropriate, a review of settings, source quality, or sampling rules. A controller should not describe footage as anonymous merely because a software workflow was used.

Accountability is established by Article 5(2) GDPR and is reinforced by several operational duties. Recital 74 states that the controller should be responsible for, and able to demonstrate, compliance with the Regulation.

  • Article 24: Controllers must implement appropriate technical and organizational measures and review and update them where necessary.
  • Article 25: Data protection by design and by default requires appropriate safeguards when selecting and configuring video processing workflows.
  • Article 30: Many controllers and processors must maintain records of processing activities.
  • Article 32: Controllers and processors must implement security measures appropriate to the risk.
  • Article 35: A Data Protection Impact Assessment may be required where planned processing is likely to result in a high risk to individuals, including certain systematic monitoring activities.

Useful interpretive sources include the Article 29 Working Party’s Opinion 05/2014 on Anonymisation Techniques (2014) and the European Data Protection Board’s Guidelines 4/2019 on Article 25 Data Protection by Design and by Default (adopted in 2020).

US equivalent

The United States has no general federal accountability principle equivalent to Article 5(2) GDPR. Privacy obligations are sectoral and state-specific. For example, the California Consumer Privacy Act, as amended by the California Privacy Rights Act, requires reasonable security procedures under California Civil Code section 1798.100(e), but it does not impose the GDPR’s general obligation to demonstrate compliance across all processing activities.

Organizations processing footage in the United States may also need to document controls under applicable contracts, security frameworks, biometric privacy laws such as the Illinois Biometric Information Privacy Act, or sector-specific rules. These obligations vary by jurisdiction and do not create a single nationwide accountability framework.

See also