104. We know that Artificial Intelligence (AI) is already being used by states as a tool to conduct information manipulation and interference. This includes the generation of AI deepfakes of politicians and other high-profile figures, including around elections, to spread disinformation and distort the truth. 105. As well as the potential for AI to be harnessed by our adversaries for malign activity, we also recognise that AI has great potential to be used positively, including in the detection, monitoring and disruption of FIMI. The FCDO continues to keep its capabilities under close review. 106. The Government has already established AI sandboxes, such as the AI Growth Lab, which has been designed as a safe space for regulators and businesses to test AI products and identify where current rules might hinder innovation. In response to the AI Action Plan, the Government committed to work with regulators to boost their capabilities. The Government has been clear that we will legislate where necessary, guided by evidence and where there are demonstrable gaps. 107. In line with the Committee’s recommendation, we will assess the existing AI sandboxes and their utility and use this to scope whether a bespoke AI counter disinformation sandbox would be useful rather than duplicative. 25