Previously, at the G7 meetings, the focus was on world leaders. Now the protagonists are Amodei, Altman and Hassabis
This week the edition was held number 52 of the G7 summit. Representatives of the seven largest world powers debated Ukraine, the Middle East, or rare earths, as expected. What was not so expected is that the true protagonists of the event were not those world leaders, but the directors of Anthropic, OpenAI, Google or Mistral who participated in the other great debate of our time: the future of AI. Power changes hands (a little). Jessica Brandt, of the Council on Foreign Relations (CFR)—an American think tank— defined the situation with a powerful phrase: “We are seeing a change in who gets a seat at the table, and a sign of where power lies.” Exceptional guests. This expert commented on how today states need to have AI companies as allies. What has happened with Anthropic and the Pentagon first and with Claude Fable 5 They then demonstrate how AI has become a weapon for country governments. One that no one wants to do without, so the leaders of the main Western AI companies have been invited to contribute to a debate that has become crucial for the future. The courtship of Amodei, Altman and Hassabis. Sam Altman (CEO of OpenAI), Dario Amodei (CEO of Anthropic) and Demis Hassabis (CEO of Google DeepMind) were the three great highlights of a group of technology executives that also included Arthur Mensch, CEO of the French startup Mistral or Alexandr Wang, head of AI at Meta. World leaders courted these managers: Narendra Modi, Prime Minister of India, met with Mensch, for example, and Amodei did it with Macronand Hassabis, Altman and the aforementioned Amodei were part of a working lunch in which Donald Trump or Emmanuel Macron were among others. Technological dependence. The situation reflects a palpable reality: the G7 leaders meet to close all types of geopolitical agreements, but they depend completely on the private infrastructure and the hardware and software of big technology. The top leaders are clear that we must get along with these companies, but the balance of power is certainly delicate: the recent US veto of Fable 5 is a demonstration. Can countries appropriate their companies’ technology and control it? AI as a weapon. Recent announcements of AI models with advanced cybersecurity capabilities such as Claude Mythos Preview and Mythos 5 or GPT-5.5 Cyber have made companies and governments noticeably concerned. Europe, for example, already complained having been set aside when Anthropic allowed some companies to use Mythos Preview. The US goes to its own. Emerson Brooking, a partner at the Atlantic Council, explained on CNBC how export controls on Anthropic models “have changed everything.” And he added that “Several G7 countries have previously alluded to the need for sovereign investments in AI, but it has always been assumed that these would take place in conjunction with access to the US technological stack. Now the US has indicated its willingness to cut off access to its AI technology both to the G7 and even to its treaties with allied powers.” Companies above governments? Advanced AI models are becoming in their own right one of the most desired and desired resources by not only companies, but also governments that are realizing what is at stake. He world geopolitical panorama already faced a similar situation with the development of nuclear weapons: only some countries can manufacture and deploy them. Digital divide on the horizon. AI is more diffuse, because models (especially open ones) are filtered, copied, and derived versions emerge from them. But one thing is certain: the data centers on which these models run are mostly under the control of companies from the US and China. Nuclear power offered deterrence, but AI can create a huge digital divide between those who control it and those who can only hope to use it. In Xataka | Claude Fable 5 has made it very clear what the big problem facing Europe is: AI is a weapon and it has none