In its Responsible AI Transparency Report, which mainly covers 2023, Microsoft touts its achievements around safely deploying AI products. Microsoft created 30 responsible AI tools, grew its responsible AI team, and required teams to measure and map risks throughout the development cycle.
Microsoft notes the addition of Content Credentials to its image generation platforms, marking photos made by an AI model. Tools for detecting problematic content and evaluating security risks for Azure AI customers were also introduced, along with new jailbreak detection methods.
Microsoft is expanding red-teaming efforts to enhance safety features in AI models, including in-house red teams and third-party testing. However, challenges have arisen due to controversies like Bing AI chatbot disseminating incorrect information and objectionable content.