Microsoft CEO Satya Nadella Calls for an Emergency Brake to Keep AI Under Human Control

Microsoft CEO Satya Nadella Calls for an Emergency Brake to Keep AI Under Human Control
Oct 11, 2026 10:35
Oct 11, 2026 21:55

Microsoft Chief Executive Officer (CEO) Satya Nadella has strongly advocated introducing an effective “emergency brake” to ensure that humans retain ultimate control over rapidly evolving artificial intelligence (AI) technology.

In a lengthy message recently published on social media, he noted that it is time to rethink the trust architecture of the technology sector. According to Nadella, highly advanced AI models, often referred to as “superintelligence,” cannot simply be treated as automated black boxes whose every decision must either be blindly accepted or outright rejected.

Separating AI Models from Their Control Systems

Under Nadella’s proposal, the core technical operations of AI models must be separated from the harnesses through which they are operated. At the same time, safety and security mechanisms should function independently, outside the models themselves.

He also emphasised the need to preserve every significant action or decision made by an AI model through conclusive evidence that is human-readable and tamper-proof.

Furthermore, the Microsoft chief believes that an emergency shutdown or pause mechanism must be available, enabling an authorised person to temporarily suspend or completely terminate an AI model’s operations at any time.

Addressing Safety Concerns and Potential Risks

According to the technology executive, every AI model should be treated as potentially risky from the outset, with stringent safety measures established accordingly.

Recently, several leading artificial intelligence companies have faced incidents involving the loss of control over their own models. Earlier, Anthropic CEO Dario Amodei also outlined a framework for developing AI with greater caution.

Against this backdrop, Nadella’s specific safety proposals have sparked fresh debate across the technology industry about how to mitigate future AI risks and ensure that humans retain ultimate authority over the technology.

//DBTech/BMT/OR//