5 Things You Should Know About Data Protection Impact Assessments under the GDPR
Privacy Impact Assessments (PIA) have long been part of Canadian public sector privacy law. However they have not been required under Canada’s private sector privacy statutes and, as a result, may be unfamiliar to many Canadian businesses and other private organizations.
The European General Data Protection Regulation (GDPR) has recently come into force, including an obligation to conduct a “Data protection impact assessment” (DPIA). DPIAs under the GDPR are similar to PIAs in Canada; however, there are significant details under the GDPR that are different.
The GDPR is a complex regulation and a complete analysis will often depend on the details of a specific circumstance. However, the following tips that may be helpful as a starting point.
1. Who needs to conduct a DPIA?
Unlike Canadian PIAs, DPIAs under the GDPR are not limited to the public sector. Under Article 35 of the GDPR, anyone who is a “controller” for the purposes of the GDPR must conduct a DPIA prior to an processing which “is likely to result in a high risk to the rights and freedoms of natural persons”.
A controller is the person or entity who “determines the purposes and means of processing”. Under Article 3, this is not limited to processing that takes place within the EU. Canadian organizations will be caught by the GDPR, and therefore subject to Article 35 (along with the rest of the GDPR) if:
- they process data about EU data subjects; AND
- the processing relates to:
- offering of goods or services (including free services) to individuals within the EU; OR
- monitoring of behaviour of individuals within the EU.
Furthermore, a new DPIA may be required for existing processing operations when the risks change, for example as a result of new technology or new uses for the personal data.
However there are some exceptions, including in cases where the data processing is carried out under a statutory obligation (and no other obligation to conduct a DPIA applies), or where the organization has recently carried out a substantially similar DPIA.
2. What is considered “high risk”?
The GDPR does not fully define what kinds of processing should be considered “high risk”. However, it does set out some specific examples for which a DPIA will always be mandatory:
- “systematic and extensive” evaluation based on automated processing, resulting in decisions which have legal effects or similarly significant effects on the individual;
- large scale processing of the “special” categories of sensitive data set out in the GDPR or of information about criminal convictions; and
- large scale and systematic monitoring of a publicly accessible area.
The recitals to the GDPR explain that the relevant risks include harms that may arise to the individual as a result of the processing, which may include discrimination, fraud, identity theft, loss of confidentiality, and loss of the data subject’s control over his or her own personal data.
The Article 29 Working Party Guidelines, together with the similar guidance from the UK Information Commissioner’s Office (ICO) make it clear that a controller must consider risks that arise from the intended use(s) of the information, as well as the potential consequences of a data breach or other unauthorized use.
Furthermore, while the GDPR is generally intended to harmonize EU law of data protection, DPIAs are one area where the data protection authorities of the EU Member States have some independent authority. In particular, each such data protection authority must publish its own lists of circumstances in which DPIAs are, or are not, required.
The ICO has recently published such a list. In the UK, DPIAs will be mandatory in a wide range of circumstances, notably including:
- any application of AI to the processing of personal data;
- any automated decision-making about an individual’s ability to access a product or service;
- any processing of biometric or genetic data (other than the use of genetic data for the delivery of primary health care);
- any combination or matching of data from multiple sources;
- any geolocation or behavioural tracking; and
- targeted marketing, profiling, or automated decision-making based on personal data of children or other vulnerable individuals, and any online services of any nature offered directly to children.
3. What is a DPIA for?
At its most basic, the purpose of a DPIA (like Canadian PIAs) is to make sure that privacy and data protection issues are considered before they arise. They are fundamentally protective measures that are intended to proactively identify and manage risks and to prevent unforeseen compliance problems.
Additionally, DPIAs (and PIAs) are themselves instrumental to establishing compliance programs.
4. What must a DPIA consider?
The minimum mandatory elements of a DPIA are:
- a comprehensive description of the proposed processing and its purposes (and, where applicable, the legitimate interest of the controller underlying the processing);
- an assessment of the necessity and proportionality of the proposed processing, in relation to its purposes;
- an identification and assessment of the relevant risks; and
- a description of the protective measures to be applied, including security safeguards and other measures sufficient to “demonstrate compliance” with the GDPR.
The ICO notes that, as part of this analysis the organization should consider:
- its lawful basis for the processing;
- how it will prevent function creep;
- how it intends to ensure data quality;
- how it intends to ensure data minimisation;
- how it intends to provide privacy information to individuals;
- how it implements and supports individuals rights;
- measures to ensure any processors comply; and
- safeguards for international transfers.
Where applicable, any relevant codes of conduct approved under Article 40 of the GDPR must be considered. Furthermore, the organization must involve its designated data protection officer in carrying out the DPIA and, should generally consider appropriate ways to obtain the views of data subjects or their representatives. While there may be circumstances in which consulting data subjects is not “appropriate”, organizations should have (and document!) a good reason if they choose not to seek any input from individuals as part of their DPIA.
5. What happens after the DPIA is complete?
If the organization concludes that the proposed processing would “result in a high risk”, it may need to consult with the competent supervisory authority in the appropriate EU Member State. In such circumstances, if the organization has not sufficiently mitigated the risks, the supervisory authority will have the authority to order changes to or to bar the proposed processing.
In any case, the organization must implement any risk mitigation or compliance measures identified in the DPIA and should document these efforts. Generally, the DPIA should be maintained as a living document as part of the organization’s compliance program. DPIAs should be reviewed periodically to ensure they remain up to date and may need to be repeated if there are substantial change to the nature, scope, context or purposes of the organization’s processing of personal data.
For more information on McCarthy Tétrault’s Cybersecurity, Privacy, and Data Management group please see our webpage.