Question to the Department for Business, Innovation, Science and Trade:
To ask His Majesty's Government what assessment they have made of the statement by Professor Yoshua Bengio that recursive self-improvement for AI has the potential to pose extinction risks and should be illegal; and what assessment they have made of the recommendation of the Joint Committee on Human Rights that the development of very powerful AI systems, which may evade effective human control, should be prohibited.
The Government takes the risks from advanced AI extremely seriously and recognises the importance of ensuring our approach keeps pace with a rapidly evolving technology. We are already taking steps to strengthen protections, including through the Cyber Security and Resilience Bill. Where further regulation, legislation, stronger protections or new powers are needed, we will act.
The AI Security Institute (AISI) was established to build a rigorous, scientific understanding of the capabilities of the most advanced AI systems and the risks they pose. It generates evidence about the capabilities of models at the frontier, and how fast they are changing, to inform national security and public safety.
The Government will respond to the recommendations in the Joint Committee on Human Rights' report, Human Rights and the Regulation of AI, in due course.