Online Harms Data Transformation

The lack of adequate data infrastructure is hindering the emergence of a competitive UK market in innovative products and services that detect and address online harms. There is an urgent need to support the development of a more mature safety tech market, in which organisations of all sizes have access to the tools they need to keep users safe online. This will help to ensure companies are able to comply with the new statutory duty of care with the lowest possible burden. The sector, however, remains at an early stage of development. Multiple interviews and consultations with stakeholders have consistently underlined that lack of access to high-quality data is among the most significant constraints to the growth of a mature and competitive market. This project will explore how best to overcome the barriers outlined above and better enable the innovative use of data to support the development of technologies that keep users safe online. It will help move us from a data landscape where data is not shared effectively to one that supports competition, innovation and improved online safety for UK citizens. It will primarily do this by appointing a provider to work with public and private sector stakeholders to gather together existing information, and explore and test innovative ways of addressing these barriers. Work will fall into two main phases: - Phase 1 - Data infrastructure discovery. To produce a report outlining the current data landscape, and the economic and social benefits of opening up online harms data. The report will set out how online harms data can be made available to a wider number of companies for the purposes of developing innovative online safety technologies. It will develop a longlist of SMART projects in which enhanced cross-sector working could have most impact, and then work with stakeholders to develop a shortlist of projects which the provider will deliver in Phase 2.

- Phase 2 - Prototype development and testing. The provider will work with stakeholders to trial and evaluate a set of practical and innovative data and social science projects, based on the shortlist developed in phase 1. This may include testing the feasibility of the two most promising solutions we are currently aware of: the use of common data taxonomies to help data interoperability, and the use of data trusts and federated learning to address issues of data sharing. The output of this work will be a series of open-source prototypes, a report setting out the findings and conclusions, and a set of practical pathways to scale successful approaches beyond the end of this project.

The provider will be required to deliver its work through a multidisciplinary project team, including deep data science and social science expertise. It will be essential to consult with a variety of UK-based and international stakeholders to deliver the project