The AI Risk Repository, created by MIT’s FutureTech project, catalogues over 700 risks identified in AI literature. This resource covers a wide range of potential dangers— from well-known risks like biased algorithms, job displacement, and misinformation to more niche concerns specific to certain programs or industries.

While many risks might be more familiar to tech or AI in education professionals, such as job displacement and privacy violations, others may not be as widely recognized. It includes two taxonomies to help users navigate risks: a causal taxonomy, which organizes risks by their underlying causes, and a domain taxonomy, which classifies them into categories like privacy, security, and misinformation.

By linking each identified risk to its original source, including paper titles and authors, the repository provides supporting evidence, such as quotes and page numbers. This comprehensive resource serves as a valuable tool for those involved in AI regulation, risk assessment, and organizational policy development, fostering a safer and more responsible approach to AI innovation.

As we explore this database, we must consider the implications of these identified risks. What new risks have you discovered that you were previously unaware of? How do you think these risks might impact your work or industry?