PIPC presents safe application of synthetic data at technology forum

[Courtesy of The Personal Information Protection Commission]

South Korea’s Personal Information Protection Commission (PIPC) and the Korea Internet & Security Agency (KISA) on Thursday held a seminar on synthetic data, a key technology in the era of a data economy.

Synthetic data refers to newly generated virtual data that only reference the characteristics of various existing information.

It differs from the concept of de-identification processing (pseudonymization or anonymization), which involves transforming some or all personal information.

When appropriately generated, synthetic data can be used without the legal restrictions required to use personal data.

As synthetic data is virtual, it can also be safely used even when sensitive information is involved, addressing concerns about personal information breaches. This makes it a notable privacy protection technology.

The key aspect of synthetic data is to maintain the utility of the actual data while ensuring that individuals cannot be identified.

One representative method is a generative adversarial neural network (GAN), which creates an artificial intelligence (AI) to generate fake data and another AI to distinguish between real and fake data, and trains the two AIs in a competitive manner to create sophisticated synthetic data.

The PIPC plans to release five reference models for synthetic data generation next week, which will cover oral images, safety helmet-wearing images, blood sugar measurement information, telecom membership usage details, and information on corporate shareholders and representatives.