As AI continues to evolve, Microsoft emphasizes the importance of trustworthy AI for organizations and communities globally. Our focus is on developing AI that is secure, safe, and private. We are committed to ensuring Trustworthy AI through leading-edge technology that safeguards our customers and developers. Today, we are unveiling new capabilities aimed at enhancing the security, safety, and privacy of AI systems: **Security**: Our top priority, reinforced by the expanded Secure Future Initiative (SFI), highlights our commitment to making our customers more secure. The first SFI Progress Report details advancements in culture, governance, technology, and operations, adhering to principles of being secure by design, by default, and through secure operations. Key announcements include: - Evaluations in Azure AI Studio for proactive risk assessments. - Microsoft 365 Copilot providing transparency in web queries to enhance admin and user understanding. Customers like Cummins and EPAM Systems have successfully adopted our security tools to bolster data protection and governance. **Safety**: Guided by our Responsible AI principles, we aim to safely build and deploy AI by avoiding harmful behaviors and risks.

We are launching new tools to help customers reap the benefits of AI while minimizing risks, including: - A real-time correction feature in Azure AI Content Safety for addressing hallucination issues. - Embedded Content Safety for on-device scenarios. - New evaluations to assess the quality of AI outputs. Customers across various sectors, including Unity and ASOS, are leveraging these capabilities to create secure AI applications. Additionally, educational institutions like New York City Public Schools and South Australia’s Department for Education are implementing safe AI chat systems for students. **Privacy**: As data is essential for AI, Microsoft is dedicated to protecting customer data through rigorous privacy principles. Today’s announcements include: - Confidential inferencing in the Azure OpenAI Service Whisper model for enhanced privacy during AI predictions. - General availability of Azure Confidential VMs equipped with NVIDIA H100 Tensor Core GPUs, which protect customer data at the GPU level. - Upcoming Azure OpenAI Data Zones will facilitate data management for AI applications in the EU and U. S. Interest in confidential computing continues to grow, with entities like F5 and the Royal Bank of Canada utilizing our secure technologies to ensure data confidentiality. In summary, Microsoft is committed to enabling trust in AI through enhanced capabilities in security, safety, and privacy. Our mission is to empower individuals and organizations worldwide to achieve more with trustworthy AI, which is central to our values of expanding opportunities, earning trust, and promoting sustainability.