In late December 2025, the New York Times published a deeply unsettling investigation with a headline that should have forced a national reckoning: Uber Cleared Violent Felons to Drive. Passengers Accused Them of Rape.
The reporting was not based on rumor or selective anecdotes. It drew from sealed court filings, internal company records, and litigation disclosures spanning years of rideshare operations in the United States. What emerged was not a story about isolated failures or bad actors slipping through the cracks. It was a portrait of a system functioning as designed.
According to the investigation, Uber approved drivers with documented histories of violent crime — including assault, domestic violence, stalking, and child abuse — and placed them in roles that involved unsupervised, one-on-one access to strangers in confined spaces. In some of those cases, passengers later accused those drivers of rape or sexual assault.
This was not a matter of outdated information or unknowable risk. The criminal histories existed. The patterns were familiar. The warning signs were not subtle.
And yet the system moved forward anyway.
If this reporting feels shocking, it is not because it reveals something we didn’t know about sexual violence. It is because it exposes how little our most visible “safety” measures are designed to confront it.
It is important to be precise about what Uber did — and did not — do.
Uber bars people on sex-offender registries from driving on its platform. Registry status is a categorical disqualifier. There is no balancing test, no individualized review, no consideration of time elapsed or demonstrated desistance. If your name appears on a registry, you are out.
This is not an accident. It is a deliberate policy choice that aligns neatly with public expectations. When people step into a rideshare vehicle, they want to believe that someone labeled a “sex offender” has been screened out. Uber can — and does — reassure riders that this is the case.
In other words, Uber complied with the most politically legible demand society makes in the name of safety. It did the thing people point to as proof that sexual violence is being taken seriously.
That fact matters, because it eliminates the easiest explanation for what followed. This was not a story about a company ignoring registries, cutting corners on the most obvious exclusions, or failing to “run the right list.”
Uber ran the list.
Uber enforced the ban.
Uber passed the compliance test.
And passengers were still harmed.
Here is the contradiction at the heart of the reporting, and the reason it cannot be dismissed as a procedural failure.
Uber excluded people on sex-offender registries.
Uber approved people with histories of violent crime.
Uber placed those drivers in unsupervised, one-on-one settings with passengers.
And sexual assaults were reported anyway.
The rideshare environment is uniquely vulnerable to abuse: confined space, minimal oversight, asymmetry of power, and social pressure to comply. If any setting demands careful attention to real-world violence risk, this is it.
Yet the system treated registry status as the decisive safety threshold — a legal label standing in for a serious assessment of who is most likely to commit harm in that environment.
The uncomfortable implication is not that Uber failed to screen hard enough. It is that the screening society demands is misaligned with the danger it claims to prevent.
If excluding people labeled “sex offenders” were meaningfully protective, this story should not exist. The registry box was checked. The reassurance was provided. The system did what it was designed to do.
What it did not do was prevent sexual violence.
The failure exposed by the reporting is not a failure to recognize danger. It is a failure to distinguish between legal labels and actual risk.
Sex-offender registry status is a legal classification. It is binary, offense-based, and largely static. Once applied, it tends to follow a person regardless of time elapsed, age, rehabilitation, or context. It tells us something about a past conviction — but very little about how likely someone is to commit sexual violence in a particular setting, under particular conditions, today.
That distinction is not academic. It goes to the heart of why the Uber screening system produced such predictable results.
Decades of criminology and victimology research have identified patterns that are far more predictive of sexual violence than registry status alone. Prior interpersonal violence. Domestic abuse. Stalking. Coercive behavior. Substance abuse combined with opportunity. Entitlement reinforced by power imbalance. These are not obscure findings. They are among the most consistent predictors in the literature.
And they were present.
The drivers approved by Uber were not blank slates. Many had documented histories of violent conduct that, taken seriously, should have triggered heightened scrutiny — especially in a context that places a single individual in control of a confined space with a stranger.
Instead, the system treated registry exclusion as the endpoint. Once the legal label was absent, the risk conversation effectively stopped.
This is the central misalignment: a static legal status was allowed to stand in for a dynamic assessment of danger, even in an environment where the conditions for abuse were well understood.
The problem does not end with Uber, and it does not begin there either.
Sex-offender registries rarely operate as stand-alone tools. They function as the moral and political anchor for a much broader regime of exclusion — one that extends into housing, employment, movement, and civic life.
Residency bans.
Proximity restrictions.
Blanket employment exclusions.
Volunteer prohibitions.
Platform bans in housing, delivery, transportation, and gig work.
These measures are often justified as necessary safeguards. Together, they create a highly visible architecture of control that signals seriousness and resolve. Communities can point to them. Companies can cite them. Policymakers can expand them.
And everyone involved can say they are doing something to prevent sexual violence.
What they rarely do is ask whether these exclusions meaningfully address the conditions under which sexual harm actually occurs.
The registry, in this sense, becomes less a prevention tool than a permission structure — a way to declare the danger identified, isolated, and removed. Once the labeled group is excluded, institutions are free to stop there.
Uber followed this script precisely. It excluded registrants. It complied with the most visible safety demand. And then it moved on, approving drivers whose histories aligned far more closely with known predictors of violence.
The pattern is familiar. Exclude the people who feel dangerous. Tolerate the people who statistically are.
This is where the concept that best explains the failure comes into focus: safety theater.
Safety theater is not about doing nothing. It is about doing things that are highly legible, emotionally satisfying, and politically safe, while leaving the underlying risk untouched.
Registries and the exclusion regimes built around them are particularly effective forms of safety theater because they offer clarity. There is a list. There is a boundary. There is a visible “other.” The public can be reassured that danger has been identified and pushed away.
What safety theater does not require is the harder work of prevention.
Prevention requires grappling with uncomfortable realities: that sexual violence is often committed by people without the most stigmatized labels; that risk is contextual and situational; that harm emerges from opportunity and power as much as pathology. It requires redesigning systems, accepting costs, slowing growth, and taking responsibility for foreseeable outcomes.
Exclusion, by contrast, requires a rule and a press statement.
Uber’s policies illustrate the appeal. By barring people on sex-offender registries, the company could credibly claim vigilance. Riders could feel reassured. The company could demonstrate compliance with public expectations — without fundamentally reconsidering how violence risk should be assessed in a rideshare environment.
The result was a system that looked strict and felt safe, while repeatedly placing passengers in harm’s way.
This is not a failure of imagination. It is the predictable outcome of a model that confuses visible exclusion with effective protection, and compliance with care.
By the time the facts in the Uber cases surfaced, there was no plausible argument left that institutions simply didn’t know better.
The risk factors associated with sexual violence in unsupervised, power-imbalanced settings have been studied for decades. They are taught in criminal justice programs, used in threat assessment, and cited regularly in public-health literature.
None of this knowledge was proprietary or experimental.
What was proprietary were the incentives.
Excluding people with violent histories narrows labor pools. It slows onboarding. It increases costs. It raises questions about liability and oversight. It requires ongoing assessment rather than one-time compliance.
Excluding people on sex-offender registries, by contrast, is simple. It is legible. It satisfies regulators and reassures customers. It creates a clear line between “safe” and “unsafe” without forcing institutions to confront more uncomfortable truths about how violence actually occurs.
This is not ignorance. It is prioritization.
Uber did not fail to imagine a safer system. It chose one that balanced risk against growth — and decided where to stop.
The Uber story forces a broader reckoning with how society allocates scrutiny.
We impose sweeping, lifelong restrictions on people labeled sex offenders — regardless of offense type, context, or demonstrated desistance. We bar them from housing, employment, public spaces, volunteer roles, and entire categories of economic participation. These exclusions are often justified as necessary for safety, even when evidence of preventive benefit is thin or nonexistent.
At the same time, we routinely reintegrate people with histories of violence — including domestic abuse, assault, and coercive behavior — once a certain amount of time has passed, or once the legal record no longer demands exclusion.
The contradiction is stark.
We restrict those who symbolize danger, while tolerating those who statistically pose higher risk in specific environments. We punish labels instead of patterns. We focus on identity rather than behavior. And then we express shock when harm occurs in exactly the conditions we failed to address.
Uber did not invent this logic. It mirrored it.
If the goal were truly to reduce sexual violence — rather than to demonstrate concern for it — prevention would look very different.
It would involve context-specific risk assessment rather than static labels. Continuous evaluation rather than lifetime exclusion. System design that accounts for opportunity, isolation, and power imbalance. Transparency about what screening can and cannot do. And accountability when foreseeable risks are ignored.
None of this is mysterious. It is simply harder than exclusion.
Prevention requires institutions to accept responsibility for the environments they create, not just the people they exclude. It requires confronting the uncomfortable reality that harm often comes from people who do not fit our most stigmatized categories.
It requires work.
The lesson of the Uber investigation is not subtle, and it does not require interpretation.
Uber barred people on sex-offender registries from driving — and still placed passengers at risk, because registry status is a legal label that provides emotional reassurance, not a serious assessment of who is most likely to commit sexual violence in an unsupervised, intimate setting.
That outcome was not accidental. It was the predictable result of a system designed to feel protective rather than to be protective.
As long as registries and broad exclusion regimes are treated as substitutes for prevention — as long as institutions are rewarded for visible action rather than effective action — this pattern will repeat. The labels will change. The platforms will change. The assurances will sound familiar.
And the harm will continue.
If we are serious about preventing sexual violence, we have to stop confusing comfort with safety — and stop accepting safety theater in place of systems that actually do the work.
- New York Times — Uber Cleared Violent Felons to Drive. Passengers Accused Them of Rape.
- Uber Background Check Policy — https://www.uber.com/newsroom/background-checks/
- CDC: Risk Factors for Sexual Violence — https://www.cdc.gov/sexual-violence/risk-factors/index.html
- WHO Fact Sheet: Violence Against Women — https://www.who.int/news-room/fact-sheets/detail/violence-against-women