Microsoft Research Advancing AI to meet needs of the global majority Learn more Featured Blogs Nov 12 MMCTAgent: Enabling multimodal reasoning over large video and image collections Tool Publication Featured Blogs Nov 11 BlueCodeAgent: A blue teaming agent enabled by automated red teaming for CodeGen AI Publication Featured Blogs Nov 6 When industry knowledge meets PIKE-RAG: The innovation behind Signify’s customer service boost Tool Publication Featured Blogs Nov 5 Magentic Marketplace: an open-source simulation environment for studying agentic markets Publication Video Featured Blogs Nov 4 RedCodeAgent: Automatic red-teaming agent against diverse code agents Publication Publications Nov 4 SigmaCollab: An Application-Driven Dataset for Physically Situated Collaboration Publications Nov 3 VCode: a Multimodal Coding Benchmark with SVG as Symbolic Visual Representation Publications Nov 4 Whisper Leak: a side-channel attack on Large Language Models Publications Nov 5 WST: Weakly Supervised Transducer for Automatic Speech Recognition Publications Nov 6 On the Coordination of Value-Maximizing Bidders Publications Nov 6 TwinVLA: Data-Efficient Bimanual Manipulation with Twin Single-Arm Vision-Language-Action Models Publications Nov 3 The Collaboration Gap Videos Oct 28 The Illusion of Inclusion How LLMs Misrepresent African Languages and Cultural Contexts Publications Nov 11 Identifying treatment response subgroups in observational time-to-event data Tools Nov 12 SPARROW Loading Opens in a new tab
Featured Blogs Nov 12 MMCTAgent: Enabling multimodal reasoning over large video and image collections Tool Publication
Featured Blogs Nov 11 BlueCodeAgent: A blue teaming agent enabled by automated red teaming for CodeGen AI Publication
Featured Blogs Nov 6 When industry knowledge meets PIKE-RAG: The innovation behind Signify’s customer service boost Tool Publication
Featured Blogs Nov 5 Magentic Marketplace: an open-source simulation environment for studying agentic markets Publication Video
Featured Blogs Nov 4 RedCodeAgent: Automatic red-teaming agent against diverse code agents Publication
Publications Nov 6 TwinVLA: Data-Efficient Bimanual Manipulation with Twin Single-Arm Vision-Language-Action Models
Videos Oct 28 The Illusion of Inclusion How LLMs Misrepresent African Languages and Cultural Contexts
