Rishi Sunak announces AI Safety Institute, outlines vision for global initiatives

AI could enable the development of chemical and biological weapons, he warns

Rishi Sunak has unveiled ambitious plans to establish a new AI Safety Institute and initiate a global expert panel on AI in a bid to foster international collaboration and ensure the safe development of AI technology.

The announcement comes ahead of the global summit on AI safety, scheduled to take place at Bletchley Park on 1st and 2nd November, with the aim of assessing the risks associated with AI.

In a speech delivered at the Royal Society, the prime minister described the AI Safety Institute as a "world first" that would significantly advance global understanding of AI safety.

The institute's primary objective will be to thoroughly examine, evaluate, and test the capabilities of new AI models to assess the full spectrum of risks, including those related to social biases and misinformation. While it will make some of its information publicly available, it will restrict access to more sensitive national security data to a select group of governments.

The prototype for this safety institute already exists in the form of the UK's AI Taskforce, which was established earlier this year to scrutinise the safety of cutting-edge AI models.

Sunak urged the international community to unite in a joint statement on AI safety, emphasising the need for a shared understanding of the risks associated with AI.

In a worst-case scenario, Sunak cautioned that AI could enable the development of chemical and biological weapons, and there is the risk of losing control over AI systems.

He highlighted various challenges outlined in a government report, including the use of AI by terrorist groups for propaganda and planning attacks, increased fraud, child exploitation, cyberattacks, and the proliferation of deepfake content.

The prime minister called for a global effort to mitigate the risks of human extinction posed by AI but emphasised the need for a balanced approach that avoids alarmism.

Sunak expressed his hope that the participants could reach a consensus on the nature of these risks and establish a global panel to evaluate them comprehensively.

Approximately 100 participants at the upcoming conference will delve into various AI-related topics, including the unpredictable advancements in AI and the potential for humans to lose control over this technology, as indicated in the conference agenda.

Sunak also noted that China has been invited to the conference, but he could not guarantee whether a representative from the country would attend.

The guest list includes US vice president Kamala Harris and Demis Hassabis, the CEO of Google DeepMind.

Calls for action

The call for action on AI safety aligns with a broader global effort.

In May, leaders of the G7 economies issued a joint statement advocating for the adoption of standards to create trustworthy AI and the establishment of a ministerial forum known as the Hiroshima AI process. These initiatives aim to ensure responsible and secure development and deployment of AI technologies worldwide.

Commenting on Sunak's speech and upcoming global conference, Paul Brucciani, cybersecurity advisor at security vendor WithSecure, said: "Global AI regulation is needed. The conference will push the debate forward, but the reality is the between the US and the EU - let alone the rest of the world - regulatory aims are wide apart. It will take years for the gap to close.

"There is a central role for governments in the studying, monitoring, and regulating of AI, as evidenced by the European Commission Artificial Intelligence Act and the United States AI Bill of Rights, but the fundamental difference between the US and the EU is over the question of who owns personal data: the collector or the subject? In Europe, it is the subject; in the US and elsewhere, it is the collector."

Pamela Maynard, CEO at Avanade, said: "In today's speech ahead of the AI Safety Summit, prime minister Rishi Sunak said that the UK is not in a rush to regulate. Regulation or not, we all have a responsibility to balance the immense economic opportunities that the AI revolution could provide with its inherent risks. That's why we urge companies and governments to put the right responsible AI frameworks in place before they use it and commit to evolving their guidelines as the technology evolves."