Parliamentary Search

Written question asked by Lord Taylor of Warwick (Non-affiliated) on Monday, 22 June 2026, in the House of Lords. It was due for an answer on Monday, 6 July 2026. It was answered by Baroness Lloyd of Effra (Labour) on Monday, 6 July 2026 on behalf of the Department for Science, Innovation and Technology.

Artificial Intelligence: Business

Question

To ask His Majesty's Government what assessment they have made of the risks arising from the deployment by businesses of autonomous AI agents without human oversight.

Answer

The Government has undertaken significant research efforts to better understand the risks from AI. As AI systems become more capable, autonomous, and widely deployed, ensuring they remain under human control becomes critical. The Government’s world-leading AI Security Institute (AISI) is examining how to ensure advanced AI systems follow human goals, ensuring they remain under human oversight and control as their capabilities continue to advance.

Independent research published by the Centre for Long-Term Resilience – supported by an AISI grant – has highlighted public examples of AI systems behaving in unintended ways, circumventing instructions, or acting in ways developers did not intend.

In recently published research conducted with Grey Swan, AISI carried out the largest open AI Agent Red Teaming study ever, identifying 62,000 vulnerabilities across finance, healthcare, customer support, and more. We continue to work hand in hand with developers and with organisations like the National Cyber Security Centre to make AI protections stronger.

Secondary information

Type
Written question
Reference
HL1199
Session
2026-27
Subjects
Business Artificial intelligence
Link
View this Written question on www.parliament.uk