Opinion: The dangers of faulty, biased, or malicious algorithms requires independent oversight.

The recent crash of a driverless car sends a clear warning about how algorithms can be deadly (1). Similarly, there are clear dangers in vital national services, such as communications, financial trading, healthcare, and transportation. These services depend on sophisticated algorithms, some relying on unpredictable artificial intelligence techniques, such as deep learning, that are increasingly embedded in complex software systems (2–4). As search algorithms, high-speed trading, medical devices, and autonomous aircraft become more widely implemented, stronger checks become necessary to prevent failures (5, 6). What might help are traditional forms of independent oversight that use knowledgeable people who have powerful tools to anticipate, monitor, and retrospectively reviewoperations of vital national services. The three forms of independent oversight that have been used in the past by industry and governments—planning oversight, continuous monitoring by knowledgeable review boards using advanced software, and a retrospective analysis of disasters—provide guidance for responsible technology leaders and concerned policy makers (7). Considering all three forms of oversight could lead topolicies that prevent inadequate designs, biased outcomes, or criminal actions. There is a long history of analyses of how poor design, unintentional bias, and malicious interventions can cause algorithms to trigger huge financial losses, promote unfair decisions, violate laws, and even cause