Trump and the agreement on a ‘spontaneous’ AI border order

The meeting at the White House between the President and the heads of the US artificial intelligence sector has resulted in a morally, but not legally, binding commitment amongst tech companies. The key points include a substantial increase in internal controls and safety systems, and an improvement in both the quality and quantity of supervisory teams.
1 OCT 26
Translated by AI
Image of Trump and the agreement on a ‘spontaneous’ AI border order

Photo: Ansa

The long table laid out at the White House, around which Donald Trump seated the leaders of American artificial intelligence – or ‘superintelligence’, to use the new term adopted by the Administration – on Tuesday, resulted in a morally, but not legally, binding commitment amongst tech companies. Everyone signed it, from Jensen Huang of Nvidia to Dario Amodei of Anthropic, including Greg Brockman representing Sam Altman and OpenAI, Mark Zuckerberg for Meta, Elon Musk with his xAI, Satya Nadella of Microsoft and Sundar Pichai of Google.
Given the purely moral nature of the pact, there are no provisions for legal coercion, nor, of course, any sanctions. It is a sort of technologically updated replica of the ‘lawless order’ – to quote a well-known book by the American jurist Robert Ellickson – which characterises the resolution of disputes between neighbours and which, to some extent, characterised the order of the Wild West. From the internal colonisation of the frontier to the frontier of artificial intelligence, in practice: after all, the agreement expressly refers to ‘responsibilities linked to the frontier’. And just as in the frontier described—with romanticised overtones—by Frederick Turner, where the arm of the law continued to operate simply by standing on the sidelines and being ready to intervene when necessary, the White House agreement also provides for the possibility that certain specific elements and areas may become subject to regulation.
Nor is it ruled out, as Trump pointed out whilst reading the clause in the text, that a committee comprising ten people representing the most significant companies could be established, headed by the White House’s Artificial Intelligence Coordinator. The key points of the agreement, which will apply to all companies involved in the AI race, consist of a substantial increase in internal controls and safety systems, particularly in the fields of cyber security and biological and chemical research, a qualitative and quantitative expansion of supervisory teams, partnerships with external and independent assessors, and the establishment of the aforementioned board, which is intended to act as a conduit and link between AI leaders, enabling them to discuss and monitor one another. The text and concept of the agreement are the result of the diplomatic efforts of David Sacks, former White House adviser on AI and cryptocurrencies, as well as a venture capitalist belonging to the ‘PayPal Mafia’ and a long-standing friend and collaborator of Peter Thiel dating back to their days at the Stanford Review. However, a substantial contribution was also made by Huang and Musk, who – not coincidentally – sat on either side of Trump at the banquet table, in true Praetorian style.
There is no doubt that the greatest problems will arise in relation to cross-checks and audits, given that all attempts to date have proved highly unsatisfactory. The very emergence of organisations such as METR – an independent evaluation body for cutting-edge AI models – has revealed decidedly opaque aspects and undeclared yet obvious conflicts of interest; several employees of the audit bodies come from the very same companies that are supposed to be audited. In reality, the agreement also implies another commitment – one that is not explicitly stated but has been discussed and agreed upon amongst tech leaders: to make AI, and above all data centres, appealing to American citizens once again. The popularity of high technology in the United States is in freefall; in March, a Gartner survey revealed that 70 per cent of respondents were opposed to the construction of data centres near their cities of residence. In mid-September, the same level of opposition was confirmed by an NPR/PBS survey. On 16 September, the House of Representatives passed the Ratepayer Protection Act, a law designed to protect energy consumers, requiring data centre owners to bear the cost of energy price rises financially without passing the burden on to the public.
The idea put forward by Trump and the Silicon Valley figures surrounding him is to establish a comprehensive system of incentives for communities that host data centres, whilst also developing technologies for infrastructure construction, heat dissipation and water consumption, in order to minimise their environmental impact.