Artificial Intelligence Legislation

Lord Tarassenko Excerpts

Monday 17th November 2025 Lords Chamber

[Baroness Lloyd of Effra (Lab)]

I thank my noble friend for her interest in this area. I can highlight that the AI Security Institute was established to provide the Government with exactly this kind of evidence and respond effectively to emerging AI risks. It has tested more than 30 frontier models, including OpenAI, Google DeepMind, Anthropic and others, and works closely with security experts across government, including the National Cyber Security Centre and Defence Science and Technology Laboratory. This is to ensure the institute’s work informs the preparations against AI-related incidents. We are committed to ensuring the UK is prepared for the changes AI will bring, and the institute’s research will continue to inform our approach.

[Lord Tarassenko (CB)]

My Lords, several European countries, including Italy, the Netherlands and Switzerland, have recently launched their own sovereign large language models to reduce dependence on models from the US and China. Now that hardware compute capabilities are no longer an issue in this country, is it not time for the UK to start developing its own sovereign large language model?

[Baroness Lloyd of Effra (Lab)]

The noble Lord asks a very good question about our sovereign capabilities. The Sovereign AI Unit’s remit spans the full AI stack, including large language models. Our priority is to secure UK access to the best models, including by deepening strategic partnerships and remaining open to backing UK companies to compete. However, we are focusing our efforts where there is greater opportunity for the UK to advance its strategic position in AI, looking across the value chain. This could mean supporting companies developing narrow models in high-impact sectors in which the UK has strengths, such as defence or drug discovery, or backing paradigm-shifting approaches in computing that can outperform incumbents.