The first global AI safety summit officially kicked off Wednesday in the United Kingdom, gathering together world leaders and tech founders to discuss the rise of artificial intelligence and address growing fears about the technology, especially so-called “frontier AI.”
US and UK Plans to Launch AI Safety Institutes
On the first day of the summit, the UK and US announced plans to set up national AI safety institutes that will focus on assessing AI models with the goal of preventing any potential harm that AI may pose.
According to reports, the UK institute will be established under the UK’s Frontier AI Taskforce, a team of AI experts set up by the government to evaluate the risk of frontier AI, which OpenAI says are “highly capable foundation models that could possess dangerous capabilities sufficient to pose severe risks to public safety.”
The US institute will be set up under the US Department of Commerce’s National Institute of Standards and Technology (NIST).
Though the institutes don’t have direct rule-making power, they will individually work to set standards for security and testing AI models in their respective jurisdictions and essentially help develop technical guidance for regulators.
Both institutes will also work closely together and exchange information on AI risks.
“We need to be working in a partnership, sharing some of our expertise, some of our intelligence, and that is certainly what the world needs to be doing. We’ve all been doing our individual work in silos. That’s got to stop now. We need to work together as well as independently,” says the UK’s technology secretary Michelle Donelan.
World Leaders Are Teaming to Tackle AI Risks
The AI safety summit ends on Thursday, and it’s expected that some other countries will put forward similar plans to develop national institutes to mitigate AI risks.
Meanwhile, the EU is planning to establish a “permanent research structure” under its EU AI Act, which the European Commission Vice President Věra Jourová says will cooperate with the UK and US institutes to prevent duplicities.
Additionally, the AI expertise centres launched in France and Canada under the banner of the French-led Global Partnership on Artificial Intelligence (GPAI) also plan to collaborate with the UK and US institutes, according to Jean Noël Barrot, France’s junior minister in charge of digital issues.
