Economic Sovereignty and the Question of Post-Deployment Training

AI systems are becoming increasingly embedded in enterprise software, agentic tools, and digital workflows. While the full labour market impacts of this tools remains to be seen, what we know from history is that access to data defines improvements in capability. As foundation models make inferences about the methods, processes and activities within workplaces easier to interpret, workplace data has emerged as a strategic economic asset. This includes not only personal data, but also tacit and explicit, individual and collective know-how: the processes, decisions, methods, and expertise that underpin firm workflows, specialisms and competitiveness.

This first release of research that is ongoing - kindly funded by Friends Provident Foundation, with support from the ESRC Centre for Digital Futures at Work - presents early findings that foundation model providers and integrated Software as a Service (SaaS) platforms may be capturing valuable business intelligence through ongoing interactions with enterprise users, enabling the codification of firm-specific knowledge that can later support automation, market concentration, and competitive displacement. In many cases, UK businesses remain insufficiently aware of the economic value of this data or the risks to their own organisational success and longevity associated with the extraction and reuse of this valuable data.

The report argues that existing governance frameworks – including data protection, human rights, employment law, intellectual property, trade secrets, competition policy, and trade law – offer important but fragmented protections. No single regime adequately addresses how workplace knowledge is captured, processed, and redistributed across the AI value chain. As a result, economic sovereignty must be understood not solely as national technological capability, but as a multi-level governance challenge involving workers, firms, markets, and the state.

Three core implications are emerging from this initial work, which warrant deeper investigation:

• Value currently created in UK workplaces risks being extracted and offshored, with UK firms often insufficiently aware of the value of their organisational know-how and how it’s being used.
• No single regime across existing governance frameworks adequately captures how workplace knowledge is captured, processed and redistributed across the AI value chain, indicating the need for real-world, socio-technical experimentation environments.
• Future regulation and industrial strategy designed to build sovereign AI capabilities must be organised around the AI value chain, combining legal, institutional, and industrial policy measures.

Ultimately, the UK’s long-term economic resilience and prosperity will depend on whether British firms and workers can retain meaningful control over the value generated from their workplaces in an increasingly AI-mediated economy.

This work was supported by the UKRI Economic and Social Research Council [grant number ES/Z504713/1] as part of the ESRC Centre for Digital Futures at Work.