Mandatory independent bias audit protocols and statutory employment agency liability for third-party AI software screening job applicants on video, voice, and resume models.
↳ Statutory Hook: NYC Local Law 144 & Title VII (42 U.S.C. § 2000e-2)Seminal state supreme court decision establishing constitutional transparency and warning constraints on proprietary machine learning risk assessment tools.
Sentencing courts may consult proprietary algorithmic risk scores only with mandatory written disclosures detailing the algorithm's black-box trade secrets, known racial disparities, and group statistical limitations.
Establishes a constitutional barrier against fully autonomous, unexplained algorithmic decision-making in public administrative and judicial proceedings.
Influential State Supreme Court Precedent (Cert. Denied)
“The use of an algorithmic risk assessment at sentencing does not violate due process, provided that the risk score is not used to determine whether to incarcerate or the severity of the sentence, and is accompanied by mandatory judicial warnings regarding the tool's proprietary nature, racial disparities, and group-level statistical limitations.”
Eric Loomis was charged in connection with a drive-by shooting in La Crosse, Wisconsin. During sentencing, the state trial court considered a Presentence Investigation Report that included a risk assessment generated by COMPAS, a proprietary closed-source machine learning algorithm created by Northpointe (now Equivant). The algorithm calculated that Loomis presented a 'high risk' of recidivism. Loomis could not review or challenge the proprietary weighting factors, trade-secret training data, or model weights.
On appeal from the Circuit Court for La Crosse County, which used a proprietary risk assessment tool (COMPAS) in sentencing defendant Eric Loomis to six years in prison.
Issue: Whether a sentencing court's use of a closed-source, proprietary algorithmic risk assessment tool (COMPAS) violates a criminal defendant's constitutional right to due process when the trade secret prevents inspection of the algorithm's methodology.
State v. Loomis is the premier American precedent establishing the doctrine of 'Algorithmic Due Process'. It defines the constitutional limits of deploying closed-source, trade-secret machine learning models in high-stakes civic decisions, establishing mandatory algorithmic transparency warnings that influenced the EU AI Act and state AI governance frameworks.
State v. Loomis, 881 N.W.2d 749 (Wis. 2016).