The CEO of Microsoft, Satya Nadellahas intervened in the debate on the pace of development of artificial intelligence, establishing a condition to advance towards superintelligence. If these systems they do not help humanity and do not remain under human control, they are ‘not worth developing’. The term ‘superintelligence’ refers to an AI that would far exceed human intellectual capabilities.
His statements come after the essay published last Saturday by Dario AmodeiCEO of Anthropic. The person responsible for Claude has asked to slow down the development of the most advanced models to give time for security measures to reach their capabilities. Their proposal includes independent evaluators within laboratories and agreements between companies and governments.
The CEO of OpenAI, Sam Altmansupports the approach and has announced that it will also allow access to external evaluators. Elon MuskCEO of xAIresponded in X that ‘Dario is right‘. For its part, Demis Hassabisco-founder and president of Google DeepMindalso supports the direction of the proposal, although it points out that details remain to be resolved.
Nadella’s position combines that support for prudence with a call for accelerate the spread of the benefits of AI. The CEO of Microsoft defends an environment in which they can Thrive both closed and open source models.
Any pursuit of superintelligence has to be grounded in the core principle that if the AI we build is not helping humanity and under human control, it’s not worth pursuing.
We also need to accelerate and spread the benefits of AI, such that they are diffused broadly across…
— Satya Nadella (@satyanadella) September 13, 2026
The manager considers it necessary to dedicate time and research to ensure that these systems act in accordance with human objectives. It has also favorably received the proposal to incorporate evaluators into the development teams, although it insists that decisions must include representatives from different countries and disciplines, including the academic field.
Microsoft plans to publish a code of Conduct for their AI models and collect opinions and improvement proposals from people and organizations. Nadella maintains that organizations must retain control of their knowledge and be able to incorporate it into models that manage, without being tied to a single supplier.
The current call is reminiscent of the letter promoted in March 2023 by the Future of Life Institutebacked by Musk and other signatories such as Steve Wozniak. They asked to suspend for at least six months the training of more powerful systems than GPT-4. The pause did not occur, although the initiative expanded the debate about the risks of AI and its regulation.
The competition with ChinaFurthermore, it makes any agreement to reduce the pace of development difficult. Amodei warns that Western containment without equivalent commitments to Beijing could give it a geopolitical advantage. That is why it proposes international coordination, in addition to internal controls.