Microsoft announced a series of artificial intelligence security, privacy and transparency capabilities, including a new feature in its Copilot AI chatbot designed to give more visibility into the components of AI responses.
The new “web search query citations” feature, expected to be generally available in October, “will include the exact web search queries derived from the user’s prompt in the linked citation section of the Copilot response,” according to a Microsoft blog post on Tuesday explaining the new feature.
It will be available for both consumer and business (Microsoft 365) versions of Copilot. The AI chatbot already shows the underlying sites that it uses to provide responses but the addition of exact search terms will be new.
Microsoft also announced several AI features for developers and administrators. Among them:
- A new correction capability in Microsoft Azure AI to fix hallucinations in real-time before customers see them.
- New simulation and evaluation tools in Azure AI studio to assess and reduce generative AI risks.
- General availability of a feature that lets customers secure data directly on Azure virtual machines that use NVIDIA H100 Tensor Core GPUs.
- New capabilities for protecting and detecting intellectual property in AI-generated code.
Microsoft is attempting to differentiate itself among AI application developers and corporate AI users against a wide range of competitors, including Amazon, Salesforce, Google, and others.
The announcements come a day after Microsoft released a progress report on its broader security initiatives. They coincide with an international “Microsoft AI Tour” that starts in Mexico City today.