‘If it’s not under human control, it’s not worth pursuing’: Satya Nadella's warning on AI

As leading AI companies race to develop increasingly powerful systems, Microsoft Chairman and CEO Satya Nadella has called for a different priority: ensuring that artificial intelligence remains under human control, delivers broad-based benefits and does not place critical knowledge in the hands of a few technology providers. 

In a post on X (formally twitter), Nadella said the pursuit of superintelligence must be grounded in a fundamental principle — if AI is not helping humanity and remains beyond human control, it is not worth pursuing. His comments come amid a growing debate within the technology industry over whether the race to build frontier AI is moving faster than safety measures can keep up. 

Nadella’s vision: AI that benefits everyone 

Nadella argued that the development of advanced AI should not come at the expense of accessibility or competition. Supporting efforts to make AI safer, he said the industry must also accelerate the spread of its benefits across countries, communities and companies. 

MUST READ | AI safety warning: Sam Altman, Elon Musk back Anthropic CEO’s call to slow down frontier AI. Why?

“We also need to accelerate and spread the benefits of AI, such that they are diffused broadly across countries, communities, and companies,” he wrote. 

For Nadella, this requires a frontier AI ecosystem in which both closed and open-source models can thrive. Such an approach, he suggested, would offer businesses and users greater choice while reducing dependence on a small number of powerful AI developers. 

His remarks reflect a broader concern about the concentration of AI capabilities, infrastructure and decision-making among a handful of companies. As businesses increasingly integrate AI into their operations, the question of who controls the underlying models and data is becoming central to the technology's future. 

Why businesses must control their own AI knowledge 

A key part of Nadella’s message was the need for organisations to retain control over their unique and tacit knowledge — the expertise, processes and insights that distinguish one business from another. 

He said every organisation should be able to build its own “continuous learning loop” or “hill climbing machine” without becoming dependent on a single model provider. 

DON'T MISS | 'Strong chance that we could all die': Ex-Anthropic AI researcher warns rapid progress could threaten humanity

In practical terms, this means businesses should be able to continuously improve their AI systems using their own data and feedback, while retaining the ability to embed their knowledge into models and weights they control. 

Nadella's position places enterprise independence alongside AI safety. Businesses, he argued, should not merely consume AI services; they should have meaningful control over how their knowledge is used, refined and incorporated into their models. 

Deliberate pacing & independent AI evaluators 

Nadella also welcomed research and deliberate pacing aimed at getting AI alignment right. Alignment refers to ensuring that AI systems behave in ways consistent with human goals, values and safety requirements. 

He expressed support for ideas such as “embedded evaluators” and wider efforts to develop mechanisms that can turn AI safety commitments into practical safeguards. 

FIND WHY | OpenAI may hit the brakes on AI development, Sam Altman tells staff

The comments came shortly after Anthropic CEO Dario Amodei urged AI companies to slow the development of frontier models. Amodei argued that the industry needs more time to strengthen safeguards and understand the risks associated with increasingly capable and autonomous systems. 

His proposals included giving independent safety evaluators permanent, employee-level access to assess AI systems, coordinating safety standards among leading AI companies and encouraging international cooperation. 

Sam Altman, Elon Musk back Anthropic’s call 

Amodei’s appeal received support from OpenAI CEO Sam Altman and xAI founder Elon Musk, two prominent figures involved in the race to develop advanced AI. 

Altman agreed that the industry should “pace the frontier” and supported the idea of independent evaluators having employee-like access to AI companies’ systems. Musk, meanwhile, offered a brief endorsement of Amodei’s position, writing on X: “Dario is right.” 

READ NOW | Are teenagers getting weak in studies? Nithin Kamath flags worrying trend as AI enters classroom

The backing from Altman and Musk is notable because their companies are competing in the same frontier AI race that Amodei has urged the industry to approach more cautiously. Their responses underscore the growing recognition that advances in AI capabilities must be accompanied by stronger oversight and safety mechanisms. 

Nadella warns against AI control by a few entities 

While supporting deliberate pacing, Nadella made clear that AI governance cannot be limited to a small group of technology companies. 

“The key is that this cannot be controlled by a handful of entities,” he wrote, calling for broad representation across the AI ecosystem, countries and fields, including academia. 

His position combines three objectives: keeping advanced AI under human control, ensuring that its benefits are widely distributed and preventing excessive concentration of technological power. 

DO CHECKOUT | 'Fully capable of influencing opinion': FM Nirmala Sitharaman raises alarm over Agentic AI

Nadella said Microsoft’s approach would focus on broad access and choice at every layer of the AI stack, enterprise control over learning loops and models, and a “Code of Conduct” underpinning the company’s own first-party MAI models. He also said the code would be published for public consultation. 

The growing discussion around frontier AI reflects a fundamental tension in the technology industry. Companies want to move quickly to gain an advantage in a rapidly evolving market, while researchers and safety advocates warn that increasingly capable systems could create risks that are difficult to manage after deployment.