Michu AI

Michu AI is a developing multilingual platform focused on AI accountability. It aims to capture and analyze real-world experiences of harmful, misleading, biased, or unexpected AI behavior, helping turn scattered reports into structured evidence. The platform is designed to connect individual experiences, reveal recurring failures and emerging risks, and support a feedback loop between people affected by AI and those working to improve it.

Users can submit reports, with options for written accounts and, where supported, screenshots, images, audio, or video evidence. Michu AI describes a process for preparing reports for analysis through language and content processing, classification, and key-signal extraction. It then aims to identify related reports using similarity analysis, clustering, and anomaly analysis, and develop findings that may inform further investigation.

The intended audiences include AI users and communities, developers and companies, safety and governance teams, researchers, and policy experts. Michu AI presents its work as complementary to benchmarks and controlled evaluations, with privacy-conscious data handling and multilingual access among its stated priorities.