Microsoft introduced ASSERT, an open-source framework designed to simplify testing AI systems for specific product behaviors. ASSERT converts natural language descriptions of AI goals and policies into detailed test cases, runs these tests, and scores the AI's performance, helping developers ensure their AI behaves as intended. This tool addresses the need for application-specific AI evaluation beyond general model benchmarks and supports continuous monitoring post-deployment. It reflects a broader industry trend toward rigorous, repeatable AI testing to improve trust and reliability.
Mayo Clinic and Microsoft announced a strategic partnership to develop a specialized AI model for healthcare. This model will leverage Mayo Clinic's clinical expertise and data combined with Microsoft's AI and cloud technology to improve diagnosis, p...
Telmai has introduced a new AI-powered data reliability workload for Microsoft Fabric, designed to automate data quality monitoring across the Fabric ecosystem. This solution continuously checks critical datasets in Microsoft OneLake, providing real-...
At its Build developer conference, Microsoft introduced MAI-Code-1, its first AI model designed to generate source code from text prompts, aiming to compete with OpenAI, Anthropic, and Google in the AI coding market. This move allows Microsoft to lev...
Microsoft has introduced Scout, a new AI assistant integrated into Microsoft 365, based on the OpenClaw framework. Scout is designed to be an always-on, adaptive assistant that learns user preferences and automates tasks like calendar management and ...
Microsoft introduced the Agent Control Specification (ACS), an open-source standard designed to help developers set clear policies for AI agents. ACS allows defining what AI agents can or cannot do, when human approval is needed, and what actions mus...
