Source URL: https://blogs.microsoft.com/blog/2024/09/24/microsoft-trustworthy-ai-unlocking-human-potential-starts-with-trust/
Source: Microsoft Security Blog
Title: Microsoft Trustworthy AI: Unlocking human potential starts with trust
Feedly Summary: At Microsoft, we have commitments to ensuring Trustworthy AI and are building industry-leading supporting technology. Our commitments and capabilities go hand in hand to make sure our customers and developers are protected at every layer. Building on our commitments, today we are announcing new product capabilities to strengthen the security, safety and privacy of AI systems.
The post Microsoft Trustworthy AI: Unlocking human potential starts with trust appeared first on Microsoft Security Blog.
AI Summary and Description: Yes
Summary: The text outlines Microsoft’s new security measures and commitments towards building trustworthy AI systems, focusing on security, safety, and privacy. It introduces new capabilities for risk assessment, content safety, and the protection of customer data, demonstrating the company’s dedication to enhancing AI security and compliance.
Detailed Description:
The text emphasizes Microsoft’s proactive stance on ensuring Trustworthy AI through a combination of security, privacy, and safety measures.
*Key Points:*
– **Trustworthy AI Commitment**: Microsoft is focused on developing AI that is secure, safe, and private, aligning their initiatives with these core tenets.
– **Secure Future Initiative (SFI)**: This initiative is central to Microsoft’s security strategy, emphasizing commitments to culture, governance, technology, and operations to elevate security standards across their products.
– **New Capabilities Announced**:
– **Evaluations in Azure AI Studio**: Designed to support proactive risk assessments.
– **Transparency Features in Microsoft 365 Copilot**: Aimed at providing better oversight on web queries related to AI tasks.
– **Customer Usage Examples**: Companies like Cummins and EPAM Systems have implemented Microsoft solutions to enhance data governance and protection, showcasing practical applications of Microsoft’s AI security measures.
– **Safety Protocols**: Microsoft’s Responsible AI principles guide the transparent development of AI, avoiding harmful content and ensuring compliance with security protocols.
– **Capabilities to Mitigate AI Risks**:
– **Correction in Content Safety**: This feature addresses and corrects AI ‘hallucinations’ in real-time.
– **Embedded Content Safety**: Allows for on-device content safety measures when cloud connectivity is limited.
– **Protected Material Detection**: Supports developers in identifying content usage within AI applications, promoting responsible coding.
– **Privacy Initiatives**:
– **Confidential Inferencing**: Provides a secure way to process sensitive data without exposing it during model predictions. This is crucial for industries with stringent data compliance requirements, such as healthcare.
– **Azure Confidential VMs**: Protect sensitive data on NVIDIA hardware, reinforcing data confidentiality.
– **Data Zones for AI**: To better manage data residency and processing, particularly important for European and U.S. customers.
– **Customer Interest**: Growing demand from entities like F5 and RBC for confidential computing solutions demonstrates the market’s increasing prioritization of data security.
Microsoft’s emphasis on AI security, safety, and privacy illustrates an evolving landscape in AI technology where responsibility and trust are paramount. The continuous development of features aimed at enhancing these aspects is critical for professionals in security, cloud computing, and adherence to compliance frameworks within software development. This commitment not only aims to solve current security challenges but also prepares organizations for future technological implications while fostering an environment of trust.