Utah's Algorithm-Based Expungement Push Shows How Data Tools Are Reshaping Record Clearing
Published on July 14, 2026 · 4 min read
Utah is preparing to deploy a new algorithm-driven tool to identify and clear roughly half a million criminal records, marking one of the most ambitious uses of automation in any state's clean slate program to date. The project matters well beyond Utah because it shows what is possible when a state combines an existing eligibility law with modern data infrastructure — and because it points to the limits that technology alone cannot solve.
Utah passed its clean slate law several years ago, but like many states with similar statutes, it struggled with the practical task of identifying which records actually qualified. The law said certain convictions could be cleared after a waiting period, but the burden of navigating the court system fell on the person with the record. The new tool is designed to invert that model: instead of asking each person to come forward, the state will scan its own records, flag cases that meet the statutory criteria, and begin the process of sealing them automatically.
How the algorithm works in practice
The tool does not make eligibility decisions in the way a judge would. It applies the criteria written into Utah's clean slate statute — offense type, time since conviction, whether the person has any new charges, and similar factors — and produces a shortlist of records that appear to qualify. A human reviewer then confirms the shortlist before a court order is entered and the record is sealed. That layered approach is intentional. The algorithm handles the volume; the human reviewer handles the cases where the data is ambiguous or the statutory exceptions apply.
For people whose records are on the list, the practical effect is that an old conviction may stop appearing in background checks without any action on their part. They may receive a notification from the state confirming that the record has been sealed, or they may find out only when a background check for a job, an apartment, or a professional license comes back clean. For people whose records are not on the list, the system is designed to make that clearer too: the state plans to publish an eligibility lookup so that people who are not automatically cleared can see why their case did not qualify.
Why this matters beyond Utah
The Utah approach is being watched closely by other states that have passed clean slate laws but have struggled to clear records at scale. Michigan showed what was possible when it cleared roughly 1.6 million records under its 2020 law, but Michigan's program also relied on a substantial state investment in data infrastructure and court automation. Smaller states have pointed to that cost as a reason their own programs have moved slowly.
An algorithm-based approach offers a more scalable template. Once the eligibility rules are written into code, the same tool can be re-run periodically as records age into eligibility, and the marginal cost of processing each additional case is low. That is especially important for the long tail of older convictions that nobody is going to file a petition for but that still show up on background checks and continue to shape people's access to housing and employment.
The limits of automation
Technology alone is not a complete solution. Records that are misclassified, names that are misspelled, or case numbers that do not match across systems can all cause an algorithm to miss a record that should qualify. Utah's tool is being deployed in phases precisely so that those kinds of errors can be identified and corrected before the system is run at full scale. And even when a record is sealed, private background check companies need time to update their databases — a record that has been cleared by the state can still appear on a commercial report until the company refreshes its data.
There is also a deeper question about who decides what gets cleared. Utah's law defines eligibility through offense categories and waiting periods, and the algorithm applies those categories mechanically. People whose cases fall just outside the statutory window, or whose records include dismissed charges alongside convictions, may still need to petition the court directly. The algorithm is a tool for executing policy, not a substitute for the policy decisions themselves.
What to watch next
The first signals from the Utah rollout will come in the next several months as the state begins processing the initial batch of records. The questions to watch are practical ones: how quickly the courts enter orders, how the state notifies people whose records have been cleared, and how many people who expected to be on the list end up not qualifying. Each of those will tell the rest of the country something about how algorithmic expungement actually performs at scale.
For people with old Utah convictions, the most useful step right now is to watch for official communication from the Utah Department of Public Safety or the Administrative Office of the Courts. Eligibility does not guarantee clearance under the new system, and people who are unsure about their status should consult the state's lookup tool once it goes live. Automatic expungement works best when the person with the record also stays informed, because even a successful program cannot predict every case.