Since the early 1970s, the accuracy principle has been a bulwark of data protection law. It basically says that organizations using personal data must ensure that these data are reasonably correct about the individuals that they represent. The principle now appears in many data protection and information privacy statutes around the world.
The EU General Data Protection Regulation says, for instance, "Personal data shall be ... accurate and, where necessary, kept up to date; every reasonable step must be taken to ensure that personal data that are inaccurate, having regard to the purposes for which they are processed, are erased or rectified without delay."
The principle has tended to be regarded as self-explanatory and has thus avoided significant academic attention. Yet, for some types of personal data, the applicability of the principle is far from clear. This is the case for personal data in the form of opinions – or inferences – about individuals.
Some have argued the principle can only apply to personal data in the form of proven facts about individuals. They argue that only facts can be judged as accurate or inaccurate. In this reading, the accuracy principle cannot apply to opinions – which are, by definition, unproven.
We argue, however, that this position is flawed and that the accuracy principle does, and should, apply to opinions.
To provide a base for our argument, we offer a working definition for the concept of an opinion. Taking inspiration from a diverse set of sources – including the work of Luciano Floridi and Mark Taylor – we propose an opinion to be: "an assertion about an entity, built on the back of facts about that entity subjected to some interpretative framework to produce new, probable facts about the entity."
In terms of whether the accuracy principle applies to opinions, we argue that the principle must apply to opinions, as both building blocks of the definition – facts and interpretative frameworks – can be considered in terms of accuracy. Regarding facts, the situation is uncontroversial. All facts can and must be treated accurately in data protection law.
Regarding interpretative frameworks, certain interpretative frameworks are trusted to produce more accurate opinions than others. For example, in the clinical context, the diagnostic opinion of a trained doctor – using an interpretative framework based on medical school and experience – will be trusted to be more accurate than that of a layperson – using an interpretative framework based on common knowledge.
The recognition that interpretative frameworks differ in the accuracy of opinions they produce allows us to draw context-specific normative lines between interpretative frameworks in terms of accuracy: those capable of producing "adequately accurate" opinions and those incapable of producing "adequately accurate" opinions.
In terms of whether the accuracy principle should apply to opinions, we argue that the core rationale behind the principle is to protect individuals against being misrepresented – through inaccurate personal data – and the consequences of misrepresentation. This rationale also applies to opinions.
Is there a risk a strict application of the accuracy principle to opinions might disproportionately impact other rights? The answer is yes. One obvious example would be how a strict application of the accuracy principle could impact the right to freedom of expression in relation to opinions in political speech.
We think, however, that highlighting this risk as a justification to exclude the applicability of the principle to opinions would be misguided. In this regard, we would highlight that the accuracy principle does not, and never has, required strict application. Rather, the principle must be interpreted flexibly such that the requisite degree of accuracy should reflect the specifics of context.
In particular, when considering how accurate personal data must be, the rights and interests of all parties involved should be considered and weighed. Thus, if opinions with a certain margin of error in precision are appropriate in context – as in, for example, opinions in political speech – these need not fall foul of the accuracy principle.
Photo by Agê Barros on Unsplash
For a deeper dive into this issue, Hallinan and Borgesius published a new paper in International Data Privacy Law.
If you want to comment on this post, you need to login.