On 19 August 2025, the Financial Conduct Authority (FCA) published the Synthetic Data Expert Group’s Report, providing insights on assessing and mitigating common challenges associated with synthetic data use.

Background

Recognising synthetic data’s potential within the UK’s financial system, the FCA set up the Synthetic Data Expert Group (SDEG) in March 2023, bringing together 20 experts across different industries to explore synthetic data and use in financial markets.

This report builds on the considerations outlined in the SDEG’s first paper responding to key feedback from the FCA’s 2022 Call for Input.

Summary

Although the report does not offer guidance, it highlights starting points for firms to embed synthetic data into existing governance controls and sets out certain good practices summarised below:

  • Existing Governance Frameworks: In the absence of a dedicated governance framework for synthetic data, practitioners are encouraged to build on existing model risk management (MRM) and Data & AI Ethics structures. The SDEG distils nine governance principles which firms may wish to consider when developing their own approaches and identify considerations that overlap or are unique to synthetic data governance. The 9 principles mentioned were: (1) Accountability (2) safety (3) transparency (4) explainability and interpretability (5) security and privacy (6) fairness (7) agency (8) suitability and (9) continuous monitoring and improvement.
  • Assessing governance and strategic readiness: The report further mentions three key governance foundations for firms to consider before launching a synthetic data project: frameworks, roles and responsibilities, and continuous monitoring. In relation to pre-project considerations, it also mentions the importance of defining clear objectives and conducting a structured value-risk assessment.
  • Regulatory, ethical and compliance considerations: SDEG members further emphasised that the generation and use of synthetic data requires consideration of key regulatory, ethical and compliance obligations such as data protection, non-discrimination and bias, oversight, and engaging with relevant compliance. It is also suggested that firms carefully consider generation methodologies and downstream implications.
  • Generating synthetic data phase: SDEG identified three areas that are relevant during the generation phase: auditability controls and monitoring, data privacy or risks and managing bias. While these issues are not exclusive to generation, early and deliberate considerations of them can help practitioners make informed design decisions and better manage risks across the synthetic data lifecycle.
  • Using synthetic data in models: Once generated, synthetic data’s impact on models ought to be validated methodically according to the SDEG report. This can be through a quality assessment using statistical comparison or techniques like performance benchmarking and Train-Synthetic-Test-Real to assess how well synthetic data supports model objectives.

Next steps

The report explains that this is intended to be a foundation and that practitioners and regulators intend to keep working together as this area develops further.