As formerly incarcerated individuals — returning citizens — exit the prison system at the state and federal levels, a particular thought keeps coming to mind. Years ago, the fear of someone who had spent twenty or more years behind bars was largely about technology of a different kind. They came home to a world that had moved from the flip phone to the touchscreen, and that adjustment alone could feel like stepping onto another planet. Today, those same concerns have been augmented by something far larger: the arrival of artificial intelligence in our society as a whole.
The risks of AI integration are widely discussed, yet I find very little written about what may be the most consequential question of all. How will AI shape the accessibility of opportunity for the people coming out of the prison system — the ones searching for meaningful employment and stable housing? The conversation about AI ethics tends to circle around privacy, misinformation, and labor at large. The returning citizen is almost never in the frame.
That absence matters.
Where AI Has Already Arrived
AI has already taken shape in the exact places returning citizens depend on. Employers now filter candidates with automated screening tools, and a growing body of research documents the biases baked into those systems. Someone with a gap in their work history, an unconventional record, or a background flag can be sorted out before a human being ever reads their name. The screening is invisible. The rejection is silent. There is no mechanism to contest it.
The same pattern shows up in service delivery. AI now triages calls to agencies and providers, so a person trying to reach a caseworker can spend thirty minutes to an hour waiting to speak to a human — if they reach one at all. For returning citizens operating with limited phone access, limited data plans, and time-sensitive deadlines, that is not an inconvenience. It is the difference between accessing a service and losing it.
The gaps are growing. The infrastructure is being built without them.
The data centers are going up. The automated hiring platforms are expanding. The AI-driven service triage systems are being deployed across both the public and private sectors. The infrastructure of an AI-driven society is being poured around us in real time — and the people with the least access to it are the ones we have already pushed to the margins.
This Is Not Only a Question of Fairness
This is a concern that belongs front and center — and not only because it is the right thing to address. How will our communities be shaped once returning citizens are reintegrated into an age where these advancements are positioned to affect them more negatively than almost anyone else?
If meaningful employment becomes harder to secure because an algorithm screened someone out, the question is not only about fairness. It is about recidivism. It is about whether a person finds a foothold or finds themselves right back where they started. Employment is one of the strongest predictors of successful reintegration. When an automated system removes that possibility before a conversation ever happens, the downstream effects are not abstract. They show up in communities, in families, and in recidivism data.
At the current pace of AI adoption across both the public and private sectors, this is not a distant hypothetical. It is the world people are walking into this year.
A Perspective From the Inside
I do not raise this from the outside. My perspective comes from having worked in the criminal justice system at the state and federal level and co-creating a reentry court program. I come to think of it as one of the ultimate tests of operational infrastructure.
Standing up a program like that means coordinating stakeholders at the federal level who each carry their own mandate, constraints, and their own definition of success. It means creating something that delivers real value while giving every stakeholder a genuine voice in how it runs. It means operationalizing that vision through coordinated effort, then holding it accountable with KPIs and OKRs so that good intentions translate into measurable outcomes for actual people. Coordination at that scale is hard, and doing it well teaches you exactly how fragile the path home can be.
That experience is part of why I take this question seriously. I have seen what it takes to build a system that actually serves returning citizens — and I have seen what happens when the systems meant to support them fail under their own weight.
What Reentry Infrastructure Must Mean in 2026
Reentry infrastructure is precisely the kind of system that should not only be needed but implemented across the board, because it is the structure capable of meeting this moment. The institutions and systems that incarcerated these individuals have an obligation to lead their return — and in 2026, that obligation now includes artificial intelligence.
Ethical AI, viewed through the social and political reality of returning citizens, cannot stop at avoiding harm. Avoiding harm is a minimum standard, not a goal. It has to mean actively equipping people with the training and the tools so that the technology lifts them rather than leaves them behind. It means designing service delivery systems that do not penalize people for having limited access. It means building hiring tools that evaluate actual capacity rather than proxies that encode historical disadvantage. It means including returning citizens in the conversations about how these systems are built — not as an afterthought, but as stakeholders whose experience makes the design better for everyone.
AI can widen the disparity or it can help close it. The same systems that hold someone accountable on the way in should help carry them forward on the way out.
The Choice in Front of Us
The choice in front of us is whether reintegration in an AI age happens by accident or by design. Left to chance, the technology will deepen the divide — because that is what unmanaged systems do to the people already standing furthest from the center.
Built with intention, I believe reentry infrastructure should be held to the same standard as any well-designed operational system: helpful, meaningful, and mission-driven — for the people coming home into a world that has changed faster than any of us anticipated.
The question is not whether AI will affect returning citizens. It already is. The question is who is paying attention — and what they are willing to build.
Frequently Asked Questions
How does AI screening bias affect formerly incarcerated job seekers?
Automated hiring tools filter candidates before a human being ever reviews an application. A growing body of research documents that these systems encode biases against candidates with employment gaps, unconventional work histories, or background flags. For returning citizens — who are statistically more likely to have all three — this means algorithmic rejection before a person ever has the opportunity to demonstrate their qualifications. The screening is invisible, the rejection is silent, and there is no mechanism to contest it.
What is the connection between AI adoption and recidivism?
Employment is one of the strongest predictors of successful reentry and reduced recidivism. When AI systems in hiring and service delivery systematically filter out returning citizens, they remove the most critical stabilizing factors in reintegration. If an algorithm screens someone out before they reach an interview, or AI service triage extends wait times to the point that a person cannot reach a caseworker, the result is structural exclusion. Structural exclusion increases the likelihood of recidivism. At the current pace of AI adoption, this is not a theoretical risk.
How does automated service triage create barriers for returning citizens?
AI now triages calls and inquiries at many government agencies and social service providers. A person trying to reach a caseworker may spend thirty minutes to an hour navigating automated systems — sometimes without reaching a human at all. For returning citizens often operating with limited phone access, limited data plans, and time-sensitive deadlines, that barrier is the difference between accessing a service and losing it. The design of these systems almost never accounts for the specific circumstances of returning citizens.
What should ethical AI look like for reentry and criminal justice systems?
Ethical AI in a reentry context cannot stop at avoiding harm. It has to mean actively designing systems that do not penalize background flags in ways that are disproportionate and unexamined. It means building service-delivery tools that preserve access for people with limited resources. It means training returning citizens on the AI tools that govern their access to opportunity, so the technology can lift them rather than leave them behind. The institutions and systems that incarcerated these individuals have an obligation to lead their return — and in 2026, that obligation includes artificial intelligence.
What is reentry infrastructure and why does it matter in an AI age?
Reentry infrastructure refers to the coordinated systems, programs, stakeholder relationships, and accountability structures that support an individual's successful return to the community after incarceration. In an AI age, it must also include a technology component: understanding which AI systems returning citizens will encounter, how to navigate them, and how to advocate when those systems produce unjust outcomes. Infrastructure that does not account for the AI landscape returning citizens walk into is already incomplete.
Why is the returning citizen rarely included in AI ethics conversations?
The AI ethics conversation has centered largely on privacy, misinformation, and labor displacement at large. Returning citizens — as a specific population navigating employment screening, housing access, and service delivery — are rarely named in that conversation. This reflects who is in the room when policy and product decisions are made, and who is not. The population most likely to be harmed by unexamined bias in automated systems is the one with the least voice in designing those systems. Closing that gap requires deliberate inclusion of returning citizens in AI ethics frameworks, policy discussions, and the design of tools that affect their lives.