Thank you for contacting me about the ControlAI campaign.
I have previously met ControlAI and discussed both the opportunities and risks of AI, including the UK’s role in international safety agreements and concerns about the Government’s current approach to copyright and regulation.
This issue is not straightforward. At the more extreme end of the risk spectrum, where Control AI focus its attention, there is a risk that western democracies are beaten to superintelligence by malevolent powers that may choose to exercise it against us. However, I agree with you that we must have the risks advanced AI may pose at the forefront of our minds.
The previous Government left behind a strong legacy on AI. Former Prime Minister Rishi Sunak was determined that the UK would be an AI superpower and the "geographical home" of new safety rules. He led the world on AI safety by hosting the world’s first AI Safety Summit - bringing industry, academia, and civil society together with 28 leading AI nations and the EU to sign the Bletchley Declaration. This landmark agreement paved the way for a new global effort in establishing a shared understanding of the opportunities and risks posed by frontier AI.
Rishi Sunak also oversaw the development of the AI Safety Institute, a global hub and international standard bearer on AI safety, which leads on vital research into the capabilities and risks of this fast-moving technology. The Institute carefully tests new types of frontier AI before and after they are released to address the potentially harmful capabilities of AI models, including exploring all the risks, from social harms like bias and misinformation, to the most unlikely but extreme risk, such as humanity losing control of AI completely. It now has more dedicated researchers than anywhere else in the world. Importantly, it received the backing of leading AI companies and nations, and helps policymakers across the globe in gripping the risks posed by the most advanced AI capabilities.
I am pleased Kier Starmer has described his inheritance on infrastructure for AI safety as “world-leading”. I understand he is intending to introduce legislation to place binding commitments on the companies developing the most powerful AI systems. The Government hopes that it would build on the voluntary commitments already secured at the Seoul and Bletchley AI Safety Summits by the previous Government. I am also aware that the Government intends to place the AI Safety Institute on a statutory footing, giving it the legal teeth needed to block emerging AI models that it assesses are a risk.
Thank you again for taking the time to contact me.
