British Prime Minister Rishi Sunak stated on Thursday that a new agreement would allow governments access to artificial intelligence models from eight leading tech companies to test them before the release, according to Politico.
Sunak announced the agreement at the end of the two-day Artificial Intelligence Summit held at Bletchley Park on Thursday. The deal on testing leading companies’ AI models was signed by Australia, Canada, France, Germany, Italy, Japan, Korea, Singapore, the US and the UK.
Until now the only people testing the safety of new AI models have been the very companies developing it. That must change.
The Prime Minister announced that eight companies, Amazon Web Services, Anthropic, Google, Google DeepMind, Inflection AI, Meta, Microsoft, Mistral AI and Open AI, have agreed to expand the access already granted to his Frontier AI Taskforce.
Access is now voluntary, although the US government has established mandatory requirements under its Executive Order to hand over certain security data.
Sunak also revealed details on the establishment of an international advisory panel devoted to the border risks of AI. It will be formed from representatives of the 28 countries that participated in the summit.
The British government stated that it would provide secretariat support for the project.
The panel will also support academic Yoshua Bengio in preparing a “State of the Science” report on the risks and opportunities of advanced AI. The paper will be published ahead of the next security summit in South Korea in the first half of the next year.
