Microsoft CEO Stresses AI Must Stay Under Human Control, Allow Teams Time to Adapt

Satya Nadella, Microsoft’s CEO, shared an internal memo on the company’s messaging platform, addressing concerns about whether the AI sector is advancing too rapidly.
He emphasized that artificial intelligence must remain under human oversight, advocated for broader third-party evaluations, and cautioned that the company must prioritize safety to maintain its operational license. "As risks escalate, invest the necessary time," he noted. "Failure to do so will result in the loss of your operating license, regardless of other factors."
Industry Pushback and Microsoft’s 37-Page Code of Conduct
Last week, several AI researchers warned that unchecked AI development could pose existential threats. Over the weekend, Anthropic CEO Dario Amodei called for a slowdown in AI progress, pledging to allow external monitoring of model advancements. Similarly, OpenAI CEO Sam Altman agreed to third-party oversight and supported delaying deployments when necessary to ensure alignment and safety.
On Monday, Microsoft published a 37-page draft of its "human-centric" AI code of conduct. Mustafa Suleiman, head of Microsoft AI, stated, "The theoretical concerns we have long held are now very real."
Core Principles: Design Goals, Customer Autonomy, and Community Impact
In his memo, Nadella highlighted three critical areas. First, clear design objectives: AI must remain under human control and benefit society broadly. This involves fostering a cutting-edge ecosystem where both open-source and closed-source models can flourish, while carefully managing the pace of development. He likened this approach to rigorous research and testing prior to release, emphasizing that increased external or third-party scrutiny is beneficial. When risks rise, sufficient time must be allocated; otherwise, the company will eventually lose its operating license.
Second, regarding customers, the focus is on ensuring they retain control over their unique and proprietary data, enabling them to build continuous learning cycles without depending on any single model provider. Microsoft’s Foundry platform supports long-running agents with enterprise-grade readiness, including built-in security, safety safeguards, and FinOps capabilities.
Third, Microsoft must ensure that local communities genuinely benefit from infrastructure projects to secure permission for development. Nadella cited the Quincy data center as an example of the company’s long-term collaborative growth with local communities.
Nadella concluded that these practices span three levels: how to innovate at the forefront, how to keep customers in control, and how to ensure economic benefits are widely distributed to communities and nations. He stated that Microsoft can point to concrete actions that back up its words.
Related article
Top 1% Firms Spend $7,500 Monthly Per Employee on AI as Computing Costs Near Labor Expenses
June 10 Update: A fresh report from the Ramp AI Index reveals that U.S. corporate adoption of artificial intelligence is accelerating. Leading firms, categorized as the top 1% of "AI power users," are investing up to $7,500 per employee monthly on AI
DeepSeek V4.1 Flash Launches With Native Multimodal Vision, Outperforming Pro Version as Prices Fall
DeepSeek has officially launched the smallest model in its latest architecture series today: DeepSeek V4.1Flash. Designed as a lightweight model with native multi-modal visual understanding, it aims to deliver higher performance, faster inference, gr
Aliyun Qwen Open Platform Launches, Enabling AI on Phones, PCs, and Glasses Across Renting, Shipping, and More
As AI applications expand rapidly, Qwen has significantly upgraded its ecosystem. Launched on August 10, the Qwen Open Platform now provides broad access for mobile phones, PCs, and AI glasses to ecosystem partners and developers.In its initial phase
Related Special Topic Recommendations
Comments (0)
0/500

Satya Nadella, Microsoft’s CEO, shared an internal memo on the company’s messaging platform, addressing concerns about whether the AI sector is advancing too rapidly.
He emphasized that artificial intelligence must remain under human oversight, advocated for broader third-party evaluations, and cautioned that the company must prioritize safety to maintain its operational license. "As risks escalate, invest the necessary time," he noted. "Failure to do so will result in the loss of your operating license, regardless of other factors."
Industry Pushback and Microsoft’s 37-Page Code of Conduct
Last week, several AI researchers warned that unchecked AI development could pose existential threats. Over the weekend, Anthropic CEO Dario Amodei called for a slowdown in AI progress, pledging to allow external monitoring of model advancements. Similarly, OpenAI CEO Sam Altman agreed to third-party oversight and supported delaying deployments when necessary to ensure alignment and safety.
On Monday, Microsoft published a 37-page draft of its "human-centric" AI code of conduct. Mustafa Suleiman, head of Microsoft AI, stated, "The theoretical concerns we have long held are now very real."
Core Principles: Design Goals, Customer Autonomy, and Community Impact
In his memo, Nadella highlighted three critical areas. First, clear design objectives: AI must remain under human control and benefit society broadly. This involves fostering a cutting-edge ecosystem where both open-source and closed-source models can flourish, while carefully managing the pace of development. He likened this approach to rigorous research and testing prior to release, emphasizing that increased external or third-party scrutiny is beneficial. When risks rise, sufficient time must be allocated; otherwise, the company will eventually lose its operating license.
Second, regarding customers, the focus is on ensuring they retain control over their unique and proprietary data, enabling them to build continuous learning cycles without depending on any single model provider. Microsoft’s Foundry platform supports long-running agents with enterprise-grade readiness, including built-in security, safety safeguards, and FinOps capabilities.
Third, Microsoft must ensure that local communities genuinely benefit from infrastructure projects to secure permission for development. Nadella cited the Quincy data center as an example of the company’s long-term collaborative growth with local communities.
Nadella concluded that these practices span three levels: how to innovate at the forefront, how to keep customers in control, and how to ensure economic benefits are widely distributed to communities and nations. He stated that Microsoft can point to concrete actions that back up its words.
Top 1% Firms Spend $7,500 Monthly Per Employee on AI as Computing Costs Near Labor Expenses
June 10 Update: A fresh report from the Ramp AI Index reveals that U.S. corporate adoption of artificial intelligence is accelerating. Leading firms, categorized as the top 1% of "AI power users," are investing up to $7,500 per employee monthly on AI
DeepSeek V4.1 Flash Launches With Native Multimodal Vision, Outperforming Pro Version as Prices Fall
DeepSeek has officially launched the smallest model in its latest architecture series today: DeepSeek V4.1Flash. Designed as a lightweight model with native multi-modal visual understanding, it aims to deliver higher performance, faster inference, gr
Aliyun Qwen Open Platform Launches, Enabling AI on Phones, PCs, and Glasses Across Renting, Shipping, and More
As AI applications expand rapidly, Qwen has significantly upgraded its ecosystem. Launched on August 10, the Qwen Open Platform now provides broad access for mobile phones, PCs, and AI glasses to ecosystem partners and developers.In its initial phase





Home






