Rasa Legal's AI Automates Expungement for 10,000+ Cases, Lowering Recidivism

The challenge of automating criminal record expungement is a classic unstructured data problem: extracting meaning from inconsistent, multi-jurisdictional court documents and mapping complex legal statutes. Rasa Legal’s AI not only tackles this technical hurdle but does so in a domain where false negatives can perpetuate poverty and false positives risk legal liability—making its reported 10,000+ case scale a significant benchmark for robustness and social impact.

The narrative of Noella Sudbury, a former public defender who founded Rasa Legal, reveals a transformative junction between criminal justice reform and artificial intelligence. As detailed in a recent NPR piece, Sudbury’s journey began with starkly observed gaps in the legal system: individuals who had served their time and sought to rebuild their lives were routinely denied employment and housing solely because of lingering criminal records. Background checks, even for minor drug or retail theft offenses, triggered subjective employer biases that equated past charges with present dishonesty. This despite research indicating that people who successfully clear their records reoffend at rates lower than the general public.

“

The narrative of Noella Sudbury, a former public defender who founded Rasa Legal, reveals a transformative junction between criminal justice reform and artificial intelligence.

Sudbury’s response, starting in 2018, was to organize expungement clinics in Salt Lake County. These events drew hundreds of attendees—some traveling hours, some bringing children—yet each night lawyers would still turn people away because the manual eligibility review was painfully slow. Determining whether someone qualified for expungement under Utah’s complex state laws could take lawyers up to an hour per case, as they sifted through physical documents. Sudbury recounts a man who arrived with a rolling suitcase filled with papers, a vivid symbol of a system so paper-based that individuals were literally buried under the weight of their own records.

This inefficiency sparked the idea for Rasa Legal, a startup that leverages AI to automate the laborious triage that previously consumed legal professionals. While the exact technical architecture remains proprietary, the core function involves accessing the same data that background check companies use—digital court records—and applying state-specific expungement statutes to instantly assess eligibility. The AI effectively performs the document review and legal reasoning that a paralegal or junior attorney would do, but at a scale and speed unattainable by humans. Rasa Legal claims it has helped tens of thousands of people seal or expunge their records, a leap from the dozens per night at a manual clinic.

This development sits at the confluence of two accelerating trends. First, the legal industry’s increasing comfort with natural language processing and rule-based AI for document review, which has been proven in e-discovery and contract analysis. Second, a broad push for access to justice, where technology is seen as a way to close the gap between legal need and affordable services. Rasa Legal’s model is not dissimilar from other legal tech tools that automate form preparation or eligibility screening, but its application to criminal records—where the stakes for individuals are enormous—marks a particularly potent use case.