Background checks on dating apps

Match Group operates more than forty dating brands and controls, by most analyst estimates, between sixty and seventy percent of the paid online-dating market in the United States. This is not a competitive marketplace where a safety-focused competitor can underprice the incumbent. It is a near-monopoly with the pricing power and the data moat to set the terms of romantic introduction for an entire country. When the company decides that background checks are not worth subsidizing, that decision propagates to roughly half of all new American couples, since Pew now reports that a majority of partnerships under thirty-five began on an app. The collective infrastructure of meeting strangers has been privatized, and the private owner has chosen engagement metrics over safety engineering.

Kathryn Kosmides built Garbo on a clean premise. Background check companies had spent decades selling fear to landlords and employers while ignoring the consumer use case that mattered most: a woman about to get into a stranger's car. Garbo would charge a low flat fee, surface only violence-relevant records, and deliberately exclude charges that punished poverty rather than predicted danger. Match Group signed on, then slow-walked integration, then quietly let the partnership lapse. Garbo wound down operations in 2024. The cause of death was not user disinterest in safety. It was a platform that wanted the brand halo of offering checks without the cost of paying for them, and a user base trained to expect everything bundled into the subscription they had already paid.

American criminal records live in roughly 3,300 county courthouses, each with its own software, retention policies, and digitization status. Some counties publish dockets online in real time. Others fax paper records on request, for a fee, during business hours. National background check vendors stitch these sources together with proprietary matching algorithms and call the result a report. The seams show. A man arrested for domestic violence in one state who moves three states over can appear clean in a national check for years. A protective order issued in a rural county may never propagate to a commercial database at all. The product looks comprehensive. The infrastructure underneath is patchwork.

A background check fed only a name and a date of birth is performing a probabilistic guess about which Michael Johnson, of the thousands sharing that name, the report concerns. Vendors mitigate this with middle names, last known addresses, and Social Security number traces, but dating apps almost never collect those inputs. The result is two failure modes pointing opposite directions. A guilty man whose name is common can pass clean. An innocent man can be tagged with a stranger's record. Both failures land on the same product, and the product is sold to users who lack the forensic training to evaluate which is happening to them.

Expungement and record-sealing are supposed to be the legal mechanism by which a society lets people move past old mistakes. In practice, commercial data brokers scrape court records before they are sealed, store the snapshots indefinitely, and resell them. Background check vendors rely on these brokers. The result is that a record sealed in 2015 still shows up in a 2024 report, often without any indication that the underlying case was dismissed, sealed, or expunged. The user reads a hit and assumes guilt. The legal system's promise of redemption is silently overridden by a private database. This is a collective harm that no individual swipe decision can correct.

Criminal records over-represent the policing patterns of the last fifty years, which means they over-represent Black and Latino men relative to actual offense rates. A background check that surfaces drug possession charges, traffic stops escalated into arrests, or pretextual misdemeanors will systematically downgrade men whose only crime was being policed more heavily. Garbo's exclusion of those categories was a deliberate engineering choice to reduce that bias. Most commercial dating-app checks have no such filter. Normalizing them at scale would import the racial geometry of American policing into the romantic-introduction layer of the country, with predictable consequences for who gets matched and who does not.

The cases that drive demand for background checks are not the gray-zone ones. They are the men with multiple restraining orders, the ones banned from one platform who immediately re-register on the next with a new email. Match Group has been sued repeatedly for failing to ban users it knew had assaulted other users on its platforms. Section 230 of the Communications Decency Act has so far shielded these platforms from most liability for what users do to each other. Until that shield narrows, the platform has no economic reason to invest in the back-end work that would actually keep serial offenders off. A background check the user pays for is a way of shifting the cost of safety from the platform that profits to the woman who is at risk.

Public conversation about background checks fixates on false positives, the innocent man tagged with a stranger's record. The deeper problem is false negatives. Most intimate-partner violence is never reported, never charged, never adjudicated. The man who has terrorized three previous partners but was never arrested will pass a check clean. Users who run checks and find nothing experience a false confidence that the system has cleared the person. They lower their guard. The check has produced the opposite of safety, an unearned trust in a stranger about whom the database knew nothing. This is the structural reason checks should be one signal among many, never the verdict.

Marie Bergström, the French sociologist who has spent a decade studying dating apps, has argued that platforms compress the early stages of acquaintance into a private, accelerated, transactional encounter that bypasses the social networks which used to do the vetting. A friend of a friend was a background check of a sort. A coworker who introduced you to her brother had skin in the game. Apps replace that with strangers meeting strangers, and a background check is one attempt to substitute a database for a social fabric. The substitution is incomplete. A record tells you what a court adjudicated. It does not tell you what his sister thinks of him.

If background checks become normalized, every user becomes a permanent suspect to every other user. The dating market begins to behave like the job market, where a single old arrest can disqualify you from every opportunity, and where the path back to a clean record requires money, lawyers, and luck. Younger users, who already live under more surveillance than any previous generation, will absorb this as the price of entry. Older users, who came of age when courtship was less documented, will feel the shift more acutely. Either way, the cost falls hardest on those whose records are dirtiest for the least dangerous reasons, which in America means poor and non-white.

A protective dating-platform infrastructure would include real-time access to protective-order data across state lines, mandatory cross-platform banning of users removed for violence, transparent platform reporting of the number and type of removals each year, criminal liability for executives who knowingly retain serial offenders on the platform, and a publicly funded record-expungement system that actually clears records from commercial databases. None of these require dating apps to become the police. They require the legal infrastructure around the apps to do its job. Until that happens, a privately purchased background check is a small, useful, deeply incomplete patch over a much larger civic failure.

Given all of the above, the practical answer for a user is unglamorous. Use the check late, not early, after you have decided you are escalating. Use it as confirmation of facts the person has already told you, not as discovery. Notice when its absence is being used to manipulate you, and notice when its presence is being used to short-circuit your own slower instincts. Tell a friend where you are going, and when you expect to be back. Meet in public until you do not. Listen to the person across the table. The database knows less than the dinner does, and the dinner knows less than the next six months.

Bergström, Marie. The New Laws of Love: Online Dating and the Privatization of Intimacy. Cambridge: Polity, 2021.

Finkel, Eli J. The All-or-Nothing Marriage: How the Best Marriages Work. New York: Dutton, 2017.

Federal Trade Commission. Consumer Sentinel Network Data Book 2023. Washington, DC: FTC, 2024.

Cramer, Cassie. "Why Match Group's Garbo Partnership Failed." The Information, March 14, 2024.

Lyngaas, Sean. "How Stalkers Exploit Gaps in Dating App Verification." CNN, August 22, 2023.

Hall, Jeffrey A. "Online Dating and the Search for True Love." Journal of Social and Personal Relationships 39, no. 5 (2022): 1234-1256.

Hood, Marlowe. "The Limits of the National Criminal Background Check." Reuters Investigates, June 7, 2022.

Tashea, Jason. "Algorithmic Risk Assessment and the Justice System." ABA Journal 103, no. 4 (2017): 54-61.

Bridges, Andrew. The Section 230 Question: Platform Liability and User Safety. Stanford: Stanford Law School Press, 2023.

Druckerman, Pamela. "What American Daters Forgot About Slow Trust." The Atlantic, February 2023.

Anderson, Margo. "County Court Records and the Patchwork of American Background Checks." Journal of Criminal Justice Information 18, no. 2 (2021): 88-104.

McLaughlin, Lisa. "Garbo, Match Group, and the Cost of Safety." Wired, April 19, 2024.