Singapore Gives Data Anonymization Tips to Companies, but Warns Training, Tech Expertise Required


Singapore’s privacy regulator has released a guide on the basic concepts and methods of data anonymization for companies that don’t plan to release the data into the public domain.

The goal is to help them protect personal data and comply with Singapore’s privacy laws. 

Data anonymization is the process of removing identifiable information from personal data so that it can’t be connected to an individual. Some data privacy laws, including Singapore’s and the European Union’s upcoming new General Data Protection Regulation, encourage anonymization by exempting organizations from some requirements if they adopt the practice.

Singapore’s Personal Data Protection Act (PDPA) exempts anonymous data from its rules as long as the anonymized data poses no risk of being reversed—or re-identified.

Anonymized data is no longer considered “personal data” because the PDPA defines “personal data” as data “about an individual who can be identified.”  As a result, organizations that anonymize their data to the extent that it doesn’t risk re-identification don’t need to comply with Parts III and VI of the PDPA. Part VI covers three areas—data accuracy, data retention, and cross-border data transfers—that are very difficult and costly for companies to comply with.

But the Privacy and Data Protection Commission’s guide also warns that its techniques won’t necessarily guarantee that data can’t be re-identified. To ensure strong anonymization, and not having to sweat PDPA compliance, companies should seek training or outside services from professionals.

Levels of anonymization balance the degree to which the data is de-identified against an organization’s uses for the data. Anonymization should be done for specific purposes, so a company should identify those purposes before de-identification, according to the guidance.

Anonymization techniques differ depending on data’s intended use.

For example, character masking, in which words and numbers are individually concealed, is used for direct identifiers such as names or Social Security numbers.

But aggregation, in which specific data sets are converted to a summary of values, is appropriate for indirect identifiers, or data points like race, religion or employment status that can be combined with other data to identify someone.  

To keep up with the constantly evolving world of privacy and security sign up for the Bloomberg BNA Privacy and Security Update.