The report outlines insights, best practices, and considerations identified by the Synthetic Data Expert Group (SDEG).
Key considerations highlighted in the report include:
➡️Pre-project considerations: Weighting a project's ethical, regulatory, and compliance implications to make an educated decision on the use and impact of synthetic data;
➡️Auditability: Maintaining a clear chain of evidence through the synthetic data lifecycle to ensure processes and decisions can be validated and tracked;
➡️Privacy risks: Synthetic data may not fully mitigate these risks, and therefore, they continue to require evaluation and consideration;
➡️Managing bias: Iterative bias management is required to ensure bias risks are mitigated rather than entrenched;
➡️Data quality: Post-generation practitioners should understand the accuracy, reliability, and utility of synthetic data for downstream applications; and
➡️Impact on models: The complexity of assessing the quality of machine learning models trained using synthetic data can drastically increase depending on the structure and objectives of the data in question.
✅Firms considering the use of synthetic data in financial crime compliance should review the report to ensure use cases harness the benefits of synthetic data while identifying and minimising unintended downstream impacts.