Microsoft has published its third annual Responsible AI Transparency Report, outlining changes to its governance approach as generative and agentic AI systems become more capable and more widely deployed.
The company said it has reworked parts of its Responsible AI Standard to address new model capabilities and risks, while expanding research and evaluation work with public-sector institutions in Singapore, Australia, the United Kingdom, and the United States.
Microsoft said its current work includes AI evaluation, human-AI interaction risks, multilingual AI capabilities, safety research, shared benchmarks, user-interface design, frontier capabilities, and specialized high-risk scenarios.
Singapore remains part of Microsoft’s AI safety work
Singapore is one of the markets where Microsoft is working with government research institutions on AI safety and security. In June, TNGlobal reported that the Infocomm Media Development Authority and Microsoft were collaborating on AI evaluation, governance frameworks, research, and access to frontier models for safety testing.
That collaboration includes work involving the Singapore AI Safety Institute and other government agencies, with a focus on evaluation methods and policy considerations for responsible access to advanced models.
In its latest transparency update, Microsoft said it continues to strengthen governance for agentic systems as AI becomes capable of taking more actions across applications and workflows. The company is also examining how interfaces and human oversight affect safety outcomes.
Microsoft said it delivered responsible AI training to nearly 20,000 engineers, policymakers, and customers over the past year. It also continues to participate in broader work on benchmarks and approaches intended to make AI risk evaluation more comparable across organizations.
The report arrives as governments and technology companies try to keep governance frameworks current with rapidly evolving AI systems. Singapore has increasingly emphasized testing, evaluation, and shared responsibility between government, researchers, and technology providers.
Microsoft said it will continue to publish its learnings, progress, and unresolved challenges as its responsible AI program evolves.
Featured image: Simon Ray on Unsplash