Building an Anonymization Pipeline

Chapter 7. Safe Use

When we described the Five Safes of risk-based anonymization in Chapter 3, we described it as a framework for the safe use of data. Although we touched on them, we left the details of trust and ethics for this chapter. Reducing identifiability will help mitigate risks, making data reuse more balanced toward benefits. But while nonidentifiabile data is no longer personal, it can still be used in ways that are creepy or harmful.

Although we can describe trust in terms of privacy and data protection in general terms, we will also look at how these concepts are applied to analytical models and decision making. Since the world of analytics is complex, involving the fields of computer science, mathematics, and statistics, we also describe some of these technologies and their challenges with respect to their safe (and therefore responsible) use.

Governance plays a role in how an organization can develop and ensure the safe use of data. This will involve a blend of ethical oversight and monitoring the integrity and credibility of analytical models, as we’ll see in this chapter. There is increasing pressure from regulators to consider data ethics, given the many ways that algorithms have failed to inspire confidence or trust. Once a framework is adopted, governance principles can be embedded in organizational tools, including technology-enabled processes, thereby getting people to acknowledge responsibility. Depending on the degree of impact and sensitivity, committees ...