Microsoft's call for responsible AI use and Azure

Microsoft has made a call to its customers in Europe to use artificial intelligence (AI) in a responsible manner. This comes as the European AI regulation (AI Act) comes into effect, which brings additional obligations for users. The company emphasizes that customers are ultimately responsible for the use of systems such as Copilot and Azure OpenAI Service. This transparent approach aims to prevent misuse and comply with privacy legislation, such as the General Data Protection Regulation (GDPR).

According to Microsoft, there is a shared responsibility model that provides basic security from the provider, while it is the customer's responsibility to guide the concrete use of AI systems. This means that organizations must establish clear guidelines for data, prompts, and oversight. Special attention is given to the need for *human-in-the-loop* mechanisms, especially for decisions that impact people, to adequately intervene if necessary.

In addition, Microsoft has pointed out the existing safety layers in Azure OpenAI Service and Copilot, such as content filters and blocks against harmful prompts. While these tools mitigate risks, they are not a substitute for the organizations' own policies and evaluations. With the Responsible AI Toolbox and Azure AI Content Safety, Microsoft offers tools to ensure that AI systems are deployed safely and responsibly, but they do require an effort from the organizations.

The requirements of the AI regulation, such as transparency and risk management, imply that organizations in the Netherlands should view their AI projects as change processes. This includes mapping *use-cases*, training staff in safe use, and logging significant activities. Microsoft also emphasizes the importance of legal guidelines and collaboration with suppliers to align responsibilities with rapidly changing regulations.

Read the full article from AI Insider.