On 9 February 2023, the FCA published Feedback Statement 23/1 ‘Synthetic Data Call for Input Feedback Statement’ (FS23/1).

Introduction

In March 2022, the FCA published a Call for Input seeking feedback on the benefits and limitations of synthetic data in resolving the challenges of data sharing, the use cases where synthetic data can provide benefits for innovation and the role of the regulator in the provision of synthetic data.

Synthetic data is a privacy preserving technique that could expand opportunities for data sharing by generating statistically realistic, but ‘artificial’ data, that is readily accessible. In particular, synthetic data is useful for augmenting rare patterns and ‘edge cases’ that are scarce in real datasets, to better train models to respond to infrequent events.

The Call for Input was launched after the FCA’s Digital Sandbox pilots and TechSprint programme showed that firms faced difficulties in accessing and sharing data in financial services, particularly new market entrants. To help resolve these, the FCA has been looking at the potential for synthetic data to widen the scope of data sharing in a way that complies with privacy requirements.

Key themes

FS23/1 sets out the responses the FCA received in relation to four key areas:

• Data access and innovation: Respondents unanimously agreed that data is crucial for innovation, although there are challenges to accessing and sharing data in financial services.
• Key themes for synthetic data: Although respondents indicated that data protection regulation places specific conditions on the data they can share and access, they also reiterated the importance of consumer privacy, and that data access should have privacy built in by design at every stage of the process.
• Synthetic data use cases and the requirements to meet them: Feedback indicated fraud and anti-money laundering as a key use case for synthetic data, in part due to its ability to augment rare patterns of behaviour in a dataset.
• The role of the regulator: Respondents indicated that the regulator could perform an intermediary role in the provision of synthetic data. Feedback also strongly indicated the need for guidelines, standards and/or governance frameworks to build trust in synthetic data and encourage wider adoption.

Next steps

In early 2023, the FCA will host a joint industry-academic roundtable in partnership with the Alan Turing Institute and the Information Commissioner’s Office to understand the challenges of validating synthetic data. The FCA plans to publish a paper in the coming months outlining its key findings and next steps.

The FCA is also setting up a Synthetic Data Expert Group, which will create a framework for collaboration across industry, regulators, academia and wider civil society on issues related to synthetic data. Applications to join the group will open later in February and will be advertised on the FCA website, with the first session expected to take place in the spring.