China, the United States, the EU and around twenty countries have signed. Wednesday in the United Kingdom the Bletchley declaration for “secure” development of artificial intelligence (AI), during the first international summit on the meteoric rise of this technology. “This historic declaration marks the start of a new global effort to build public trust in AI by ensuring it is safe,” said British Prime Minister Rishi Sunak on X (formerly Twitter). The European Union and 28 countries meeting at Bletchley Park in north London agreed on “the urgent need to collectively understand and manage the potential risks” of AI through “a new global effort to ensure that AI is developed and deployed in a safe and responsible manner.” With the growing potential of models like ChatGPT, Bletchley's statement “shows that for the first time, the world is coming together to identify the problem and highlight its opportunities,” British Technology Minister Michelle Donelan told AFP. This meeting “does not aim to lay the foundations for global legislation, it must serve to chart a path forward”, she clarified. Two international summits on AI will follow, in South Korea in six months then in France in a year, she added from the emblematic center for decrypting codes from the Second World War. For two days, political leaders, AI experts and tech giants gathered there at the initiative of the United Kingdom, which wants to take the lead in global cooperation on this technology. At the same time, American Vice President Kamala Harris is due to announce in a speech in London the creation of an institute on the security of artificial intelligence in Washington. Such a structure – similar to the one that the United Kingdom has also announced to establish – would bring together experts responsible for establishing “guidelines” and evaluating the most advanced AI models to “identify and mitigate” risks, according to the White House. Generative AI, capable of producing text, sounds or images on simple request in a handful of seconds, has made exponential progress in recent years and the next generations of these models will appear by summer. They raise immense hopes for medicine or education, but could also destabilize societies, make it possible to manufacture weapons or escape human control, warned the British government. – “Behind closed doors ” – After this first day devoted to the potential dangers of the most advanced AI, high-level political representatives are expected at Bletchley Park on Thursday. 
Among them, the President of the European Commission Ursula von der Leyen, the Secretary General of the UN Antonio Guterres and the Italian Prime Minister Giorgia Meloni – the only head of state of the G7 to make the trip.
The United Kingdom hopes to convince them to create a group of experts on AI on the model of the IPCC for the climate.
The American billionaire and star entrepreneur Elon Musk, already present at the summit on Wednesday, will speak with British Prime Minister Rishi Sunak on Thursday evening.
In an open letter published Tuesday, several of the “founding fathers” of this technology such as Yoshua Bengio or Geoffrey Hinton pleaded for “the development and ratification of an international treaty on AI”, in order to reduce the risks “potentially catastrophic consequences that advanced systems pose to humanity.”
The challenge is to be able to define safeguards without hindering innovation for AI laboratories and tech giants. The EU and the US, unlike the UK, have chosen the regulatory route.
England's Minister for Science, Innovation and Technologies, at the Bletchley Artificial Intelligence Summit, November 1, 2023 © POOL – TOBY MELVILLE
Last week, several companies such as OpenAI, Meta (Facebook) or DeepMind (Google) agreed to go public some of their security rules on AI at the request of the United Kingdom.
In an open letter addressed to Rishi Sunak, around a hundred international organizations, experts and activists deplored that this summit was held behind closed doors, dominated by tech giants and with limited access for civil society.
All rights of reproduction and representation reserved. © (2023) Agence France-Presse