Microsoft boss on AI content: 'Nobody wants anything that is sloppy'
The Register
February 25, 2026
AI-Generated Deep Dive Summary
Satya Nadella, Microsoft’s CEO, emphasized the importance of trustworthiness and accuracy in AI tools during a London AI tour. He highlighted that “nobody wants anything that is sloppy” when it comes to AI creation, underscoring the need for reliable and verified outputs from AI systems like Copilot. This message was particularly relevant as Microsoft showcased its ambitious plans for AI-powered agents and their potential to transform work processes. However, the event also revealed a critical challenge: AI cannot be fully trusted, and human verification is essential to ensure accuracy.
The conference demonstrated various AI capabilities, such as data collation in Excel and automated test plan execution, but it was clear that AI’s fallibility remains a significant hurdle. For instance, a screen during a Copilot demo warned, “Copilot uses AI. Check for mistakes.” This acknowledgment of AI’s limitations aligns with Nadella’s broader focus on engineering quality and the need for an “AI sovereignty” approach—ensuring control over AI systems rather than just where they run.
While Microsoft avoided discussing the West Midlands Police’s Copilot incident, where the tool produced incorrect information about a football match, the issue of AI reliability remained central to the discussions. Nadella’s remarks highlighted the importance of addressing these challenges head-on, especially in critical areas like policing and decision-making. His call for an engineering quality czar further underlines Microsoft’s commitment to improving AI’s accuracy and trustworthiness.
The event also touched on UK-specific AI use cases, such as time-saving tools for doctors and civil servants. These examples illustrated AI’s potential benefits but also reinforced the need for human oversight. As AI
Verticals
tech
Originally published on The Register on 2/25/2026