Clean slate laws are sweeping the country, offering many of the estimated 70 to 100 million people with a criminal record the chance to have their record expunged. The benefits seem straightforward: Making a criminal record no longer publicly available should reduce housing and employment discrimination. The policy aims to give people a second chance, especially those who were unfairly targeted to begin with. Expungement has largely been framed as a way to address the errors of a legal system rooted in racial hierarchies and discrimination.

Key to this new wave of legislation is the effort to make expungement automatic by shifting the burden to the state rather than the person with the record, and automating the process as much as possible using technology. These policies are a welcome relief against onerous expungement processes, like inefficient, confusing, and expensive court petitions that caused most people to drop out of the expungement process or never try at all, creating a “second chance gap.” One study in Michigan, for instance, found that only 6.5 percent of people eligible for an expungement successfully completed the process. By making the process both automatic and automated, many hope that expungement can reach more people, particularly those who can’t afford an attorney, or who understandably do not want to reengage with the court system even for record-sealing purposes. Over a dozen states have implemented automatic record clearance, including eight states that authorize automatic relief for cannabis-related convictions.

Unfortunately, many states lack the data infrastructure necessary to effectively seal criminal records. A massive backlog in California has left tens of thousands of people who are legally eligible for record clearance still waiting. And new research in California—whose clean slate law, AB 1076, automates the expungement process for people arrested after January 2021— shows that these promising new laws may have the unintended consequence of increasing racial inequality because of how narrowly they're written. People with more serious records or repeated contacts with the criminal legal system are often excluded from clean slate policies. But the issue of who has a more serious criminal record is deeply structured by the race, neighborhood, and income of the person who is arrested and charged.

Clean Slate laws have received broad bipartisan support. But making laws politically palatable to both sides of the aisle can result in narrow policy. And while advocates point out the common sense benefits of criminal record expungement, public opinion for expungement is mixed: One recent public opinion survey found that nearly 55 percent of respondents were opposed to expungement on the grounds that having public access to criminal records “keeps communities safe.” But the same study showed that less than 15 percent of respondents felt a person should never be able to get an expungement, with support for record clearance policy rising for property and substance-related offenses and after a person has remained crime-free for seven to ten years. Both the dominant political framework and public opinion have encouraged expungement policy for only low-level, nonviolent crimes and for those who have proven they can remain crime-free.

The violent/nonviolent crime dichotomy is a bit complicated: Many states deem a broad variety of offenses as violent, including things most people would consider nonviolent, like burglary, drug crimes, and embezzlement. And “crime-free” is a more nuanced concept than we often think. For the purposes of record clearance policy, criminal behavior is often measured by new arrests or criminal charges and convictions. But living in an overpoliced neighborhood can lead to an increased likelihood of being stopped by police.

New research from California looked at over 2 million Californians who have been arrested at least once. Black people were more likely than other race groups to have been convicted of a criminal charge (87.3 percent, versus 79.4 percent of total), and of those convicted, were more likely to have a felony record (73.3 percent versus 58.1 percent of total). 40 percent of Black people are barred from expungement due to the type of crime, compared to 26 to 31 percent of people in other race groups. This means that a disproportionate share of Black people have a felony conviction that disqualifies them from expungement. Even among people with felony records in other race groups, Black people were less likely to have a conviction that fit the criteria and were much more likely to have been sentenced to prison for their conviction, rendering even more people ineligible for record clearance.

“What our study shows,” study authors Alyssa Mooney, Alissa Skog, and Amy Lerman told me in an email exchange, “is that automating record clearance alone is not going to be sufficient to reduce racial disparities in who has a criminal record. ... What will be needed to actually reduce the racial gap in criminal records is a policy change that extends record clearance eligibility to a wider range of cases. This isn’t a technology problem; it’s a political problem.”

The underlying issue is the persistence of historic racial inequalities in the American legal system. Seemingly race neutral policies can have racially disparate effects that perpetuate inequalities.

“In a country where poverty, education, geography, health, and so many other factors are as highly stratified by race as they are in America, there’s really no such thing as ‘race blind’ policy. Policies don’t start from nowhere; they start from where we are right now in history, and layer on top of all the inequalities that already exist,” note the researchers of the California study. “We know that Black Americans have historically been much more likely to live in heavily policed neighborhoods, to be stopped and questioned by police, and to be sentenced to prison or jail. That means when you pass a law that limits criminal record clearance to only people who have committed some types of crimes, or who have some types of criminal records, it is going to have a different impact across racial groups.”

Even improvements to the technology of record clearance may produce inequalities. The California analyses revealed that 35 percent of state cases were missing a final disposition in the data, and 75 percent of people had at least one such incomplete case on their record. Recent research also shows that patterns of bad data are also linked to race, where states with a disproportionately large number of people of color having felony convictions also suffer more incomplete data, stymieing reform efforts at the scale of automatic expungement. Automation cannot work when criminal record data is so incomplete.

This isn’t the first time that a legal system reform delivered unintended racial consequences. A few years ago, economists discovered a troubling trend in the wake of widely lauded “ban the box” laws that blocked employers from asking about an applicant’s criminal record until an offer of employment was made. The idea was that removing the checkboxes that inquire about a conviction record would reduce record-based discrimination. The research showed, however, that the policy might have increased racial discrimination against young Black male applicants without criminal records, because they lost the opportunity to defy employer stereotypes about race and crime by signaling that they had never been convicted of a crime.

There are concerns that sealing criminal records may have a similar effect, or might incentivize employers to use private sector background check data or simple Google searches to locate sealed information.

As Emily Baxter of the organization We Are All Criminals reminds us, “one in four people in the U.S. has a criminal record; four in four have a criminal history.” In other words, we all break the law, but only some of us get caught—and get a criminal record. When expungement policy favors those who were able to get their charges dismissed or convince a prosecutor to downgrade a charge from a felony to a misdemeanor, expungement is increasingly available to only those who had more resources or were in an advantageous position at the time of their arrest.

Clean slate laws can open doors for people who are barred from opportunities, particularly in industries that require state licensing, such as nursing. The individual impact of having one’s record sealed is overwhelmingly positive for personal, economic, social, and professional reasons. Much enthusiasm around expungement has taken a racial justice approach as well: For instance, cannabis expungement has centered on how the war on drugs disproportionately targeted people of color.

This new research, then, tells us that we might consider expanding record sealing to many more people. This would reduce the racial inequalities that may emerge from the current policy landscape and open the doors to many more people deserving of a second chance.

The study authors point to policy interventions that would help eliminate the patterns they discovered, including a seven-year sunset rule and allowing people who have pending charges or who are still on probation to have their older records sealed. The sunset rule mirrors state policy guidance for private background check companies, which bars the reporting of offenses from over seven years ago. It would also automatically clear all records from over seven years ago regardless of a person’s otherwise disqualifying criminal record. New Jersey has taken this approach with its new 10-year clean slate law, which will soon automatically clear a person’s entire criminal record after they remain crime-free for a decade, even if that person is otherwise ineligible for traditional, petition-based expungements. Many other states are expanding or considering expanding their expungement policy to allow access for more people, including those with felony convictions. This is good policy.

Automation and algorithmic approaches to expungement get a lot of attention for making the process fairer and for expanding access, but the underlying policies may be increasing race-based discrimination in the long run by cutting people of color out of the eligibility pool. The lesson is that technology alone can’t fix an inherently unfair policy, but thinking more broadly about second chances—and who deserves them—might.