Centre for Data Science & AI

Colleagues from the School of Computing Science in the College of Science & Engineering have developed machine learning models for accelerating the identification of context-dependent sensitive information - for example Freedom of Information Act (FOIA) exemptions - to assist with the sensitivity review of digital data documents in Government departments. This is critical for driving efficiency and ensuring that certain information is protected before the documents are released to the general public and has implications for maintaining the UK’s international relations and national security. However, there are vast amounts of data to sift through as this information is transferred to The National Archives, making this a challenging and time-consuming task that would benefit from Artificial Intelligence supporting human endeavour.

Soon after this project commenced, FCDO Services realised that the ideal software for this task did not exist and needed to be designed from scratch, ensuring compatibility with hardware, robustness and future-proof while being readily usable by expert sensitivity reviewers to operate. Furthermore, given the nature of the data, any system had to satisfy rigorous government security protocols.

Funded by a Knowledge Transfer Partnership Innovate UK grant, the technology is currently being trialled by the FCDO and its partners led by SVGC Ltd. and has attracted interest from other departments in Whitehall. The significance of this cutting-edge technology developed at the University of Glasgow in the context of data security was highlighted in two reports commissioned for the Prime Minister:

• Government digital records and archives review by Sir Alex Allan

The system also featured in the FCDO Services annual report in 2018/19.

---

First published: 1 January 2023