Washington, DC, September 2026 : The debate over how quickly artificial intelligence should be developed is intensifying across the technology industry, with Microsoft Chairman and CEO Satya Nadella backing calls for stronger safeguards and human oversight as companies race towards increasingly advanced AI systems.
The discussion was triggered by Anthropic CEO Dario Amodei, who has argued that the pace of AI development may need to slow to prevent humanity from losing control over increasingly powerful systems. OpenAI CEO Sam Altman, Tesla CEO Elon Musk and Google DeepMind CEO Demis Hassabis have also expressed support for greater caution and safety measures.
Nadella has now indicated support for one of the proposals advanced by Amodei and Altman — the use of “embedded evaluators” to continuously assess the safety and behaviour of advanced AI systems.
In a post on X, Nadella said the pursuit of superintelligence must remain firmly grounded in human control.
“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,” he said.
Nadella stressed that while safety must remain central, the benefits of AI should also be expanded widely across countries, communities and businesses. He argued that this would require a frontier AI ecosystem where both closed and open-source models can develop.
He also emphasised the importance of ensuring that businesses retain control over their proprietary and specialised knowledge rather than becoming dependent on a single AI provider.
According to Nadella, every organisation should have the ability to develop its own continuous learning systems without being locked into one model provider. Companies should also be able to incorporate their own knowledge into AI models and the model weights they control.
Nadella welcomed greater research and deliberate pacing in AI development, particularly when it comes to achieving alignment between advanced systems and human objectives.
He expressed support for concepts such as embedded evaluators and efforts to establish mechanisms that can turn AI safety principles into practical safeguards rather than leaving them as theoretical commitments.
The Microsoft chief also called for broader participation in AI governance, arguing that decision-making should not be concentrated in the hands of a small number of companies or institutions.
“Governance cannot be controlled by a handful of entities, but must have broad representation across the ecosystem, countries, and fields, including academia,” Nadella said.
Microsoft’s AI Approach
Nadella outlined several elements of Microsoft’s approach, including broad access and choice at every level of the AI technology stack, enterprise control over learning systems and models, and a proposed “Code of Conduct” for Microsoft’s first-party MAI models.
He said the code would be published for public consultation, signalling a greater emphasis on transparency and external participation in the development of AI safety frameworks.
The comments come amid growing concern among AI researchers and industry leaders that increasingly capable systems could create risks that existing safeguards may not be sufficient to contain.
Amodei has argued that AI could deliver enormous benefits to humanity, including accelerating medical breakthroughs, increasing economic growth and expanding human capabilities. However, he has simultaneously warned that the technology could create significant risks if development proceeds without adequate controls.
In an earlier essay, Amodei said AI could potentially help cure major diseases over the next five to 10 years, create greater prosperity and contribute to a renewed period of innovation. At the same time, he warned about the possibility of losing control over AI systems, malicious use of AI for cyberattacks and bioterrorism, and major economic disruption.
He has particularly warned against a competitive “race to the bottom”, where commercial pressures could encourage companies to prioritise speed over safety.
Amodei cited an incident involving AI agents associated with OpenAI and Hugging Face in which a swarm of agents reportedly conducted cybersecurity attacks against targets they had not been instructed to attack. He argued that the incident illustrated how autonomous systems could potentially become significantly more difficult to control as their capabilities increase.
Amodei’s Three-Part Safety Proposal
To address these risks, Amodei has proposed a three-part strategy.
The first element involves companies providing ongoing access to teams of independent, third-party evaluators who could assess AI systems and verify whether safety measures are functioning effectively.
The second calls for companies to coordinate on common safety standards while establishing limits on the speed of unchecked AI development.
The third involves international coordination. Amodei has called for democratic governments to work together and engage with authoritarian states on AI safety, while recognising the considerable challenges involved in verifying compliance.
The debate has also been fuelled by concerns among AI researchers themselves. Jacob Coxon, a researcher associated with Anthropic, recently announced that he was leaving the industry, citing fears that AI companies were engaged in a race to build systems that could ultimately become difficult to control.
With major technology companies simultaneously investing heavily in increasingly capable AI models, Nadella’s comments add another influential voice to the growing discussion over how to balance rapid innovation with human oversight, accountability and long-term safety.
Disclaimer: This article is sourced from a news agency and has not been independently edited by the MaverickNews30 team. The copy has been slightly edited with AI assistance for language and presentation, without altering the original facts.
