British Prime Minister Rishi Sunak is proposing the establishment of a new artificial intelligence (AI) body to assess the national security implications of this emerging technology. The proposed AI Safety Institute would work alongside governments to evaluate the risks associated with advanced frontier models of AI. The plan is for the institute to develop from the existing Frontier AI Taskforce, a government body currently in discussions with major AI companies Anthropic, DeepMind, and OpenAI to gain access to their models.
The UK’s Science, Innovation, and Technology Secretary, Michelle Donelan, has previously suggested that the Frontier AI Taskforce could evolve into a more permanent and potentially international institution. As part of the plan, the UK aims to create a highly secure facility where national security officials can conduct risk assessments of the frontier models.
The second day of the upcoming two-day AI summit, hosted by the UK, will bring together representatives from G7 countries, other Western allies, and executives from major AI companies. The summit will serve as a platform to discuss the implementation of voluntary commitments made by leading AI companies to manage the risks posed by their models. The goal is to reach a shared understanding of the risks and opportunities presented by frontier models.
Topics to be discussed on the first day of the summit will include the risks associated with losing control of frontier AI, such as potential threats to global safety through the misuse of AI. The conference will address concerns such as the use of AI to develop bioweapons and launch cyberattacks, as well as the risk of humanity losing control of AI technology.
The proposed AI Safety Institute and the discussions at the summit highlight the UK government’s commitment to ensuring the responsible and secure development of AI. By collaborating with like-minded countries and major AI companies, the UK aims to evaluate and address the potential risks associated with frontier models, safeguarding national security in an increasingly AI-driven world.
More detail via POLITICO here… ( Image via POLITICO )