The first global summit on AI ends with an insufficient declaration on a phenomenon full of risks
The first Global Summit on Artificial Intelligence (AI) Security took place this Wednesday and Thursday at the famous headquarters of the British Government Intelligence Code and Cipher School, the place where Alan Turing broke the encryption of the Enigma machine . All so that the 28 governments attending, including the EU, the US and China, can say that they have signed the Bletchley Declaration, where they affirm that AI has the potential to significantly improve human life and at the same time recognize that it “poses risks.” significant”, even catastrophic, for their well-being. Together they stage the need to regulate their development collectively. They did so in the presence of some of its main architects, such as Demis Hassabis (Google DeepMind), Sam Altman (OpenAI) and Elon Musk (owner of Tesla or SpaceX, co-founder of OpenAI and owner of the social network X, formerly Twitter).
The message is urgent. Sunak proposed creating an Intergovernmental Expert Group on AI, similar to the IPCC dealing with Climate Change, and a summit every six months, with upcoming stops in South Korea and France. The day before, the G-7 announced its code of ethics for companies that develop Artificial Intelligence systems and Joe Biden promulgated an executive order to promote and control AI, a technology that he declares “the most momentous” in recent history. In that order, the US president invokes the Defense Production Act to require all next-generation developers to share the results of their regular tests and audits and any critical information with the Government.
It is a law signed by Harry Truman in 1950 that grants the president authority to ensure the availability of critical resources and technologies necessary for national security. Designed for states of exception such as wars, natural disasters and pandemics, in this context it is reminiscent of China’s Cyber Security Law, which since 2017 grants the Beijing Government powers of supervision and regulation of technology companies in its country. Both put the private sector at the service of the State or, as Biden says, “maintaining the progress of American leadership globally.”
The success of this first summit was to bring together, in the middle of the chip war, the US Secretary of Commerce, Gina Raimondo, with the Chinese Vice Minister of Science and Technology, Wu Zhaohui. Rishi Sunak, host of the meeting, failed, however, to establish the first AI testing center in the United Kingdom. In fact, Raimondo announced that the US will create its own AI security center, and with its own standards. The failure is that the Bletchley Declaration is not binding. Anyone can sign it because it does not obligate anything. To begin with, most of the measures promoted by Biden require the approval of the US Congress, which is absolutely polarized.
AI must be regulated, but intention is not enough. Negotiations on the Artificial Intelligence Law continue in the European Parliament under the Spanish presidency. This regulation does aspire to establish the legal framework for the development of safe AI. The Minister of Economic Affairs and Digital Transformation, Nadia Calviño, has stated that it could be signed in three weeks. It is a reasonable period. Full of advantages but also risks, such as the disappearance of jobs or large-scale misinformation, Artificial Intelligence poses challenges every day that do not wait for any law.