This Article proposes a new role for legal algorithms as a tool for critiquing the law. To date, algorithms have been used as a tool for improving how legal predictions are made, typically by reducing the influence of human bias and human error. These algorithms have come under fire, particularly in the domain of criminal law, where they have been shown to produce racially biased and inequitable outcomes. But oftentimes legal algorithms are not independently causing harm; they are instead revealing the harm of the laws that they apply. In fact, machine learning algorithms have a hidden talent for exposing how legal predictions work and what outcomes they can produce. This Article coins the phrase “algorithmic exposure” to describe how basic machine learning techniques can be used to assess a law’s efficacy and fairness. Through this process, legal algorithms can provide empirical support for critiques of law by revealing how a law falls short of its goals or systematically burdens certain groups. Algorithmic exposure should be adopted as a routine, best practice for assessing laws that involve prediction. This Article uses pretrial incarceration doctrine as a lens for surveying how algorithmic exposure could be incorporated within political advocacy, legislation, and agency policy. Advocates could galvanize the public by using algorithms to reveal how contemporary pretrial incarceration laws are ineffective and racially inequitable, no matter how they are applied. Likewise, legislatures could use algorithms as a diagnostic tool for calibrating potential legislative reforms. Independent of statutory changes, prosecutors could use algorithmic analysis of pretrial doctrine to justify, both in court and to the public, internal policy reforms that reduce pretrial incarceration. Taken together, these examples reveal how legal algorithms could become an unexpected ally in efforts to transform our legal systems.