Can we harness AI’s potential to expand access to safety and justice without repeating the harms it has already caused?
A few years ago, I was talking with a foundation about funding our work addressing the extraordinary overrepresentation of people with disabilities in the criminal legal system.
The foundation liked our proposal. But then came a request: Could we add an AI component?
I'll admit, I was irritated.
We were trying to address a deeply entrenched injustice. Did we really need to incorporate the latest technology trend to get funding?
But the request stayed with me. And eventually, my irritation gave way to curiosity.
What if AI could actually help us solve some of the problems we've been working on for decades?
The more I thought about it, the more I realized there was a fascinating—and troubling—contradiction at the heart of that question.
The promise and peril of AI
For people with disabilities, AI has already been a game changer.
AI-powered tools can describe surroundings and read documents for people who are blind. Real-time captioning can make conversations more accessible for people who are hard of hearing. Advances in speech recognition and communication technology are creating new opportunities for people with speech disabilities. And AI can help people with intellectual and cognitive disabilities understand complex information and navigate everyday tasks.
These innovations are doing more than making life convenient. They're expanding independence, autonomy, and access.
But in the criminal legal system, the story has been profoundly different.
AI and algorithmic technologies have been used to expand surveillance, predict where crime might occur, and assess people's risk of reoffending. These tools have too often reproduced or amplified the racial inequities embedded in our criminal legal system, with devastating consequences for individuals and communities, particularly communities of color.
Think about that contradiction. The same technological revolution that is expanding freedom and independence for some people is reinforcing surveillance, discrimination, and control over others.
And at Activating Change, we work at the intersection of these two realities.
What if we could harness the promise without repeating the harm?
People with disabilities experience violent victimization at approximately three times the rate of people without disabilities. They are also dramatically overrepresented in our nation's jails and prisons.
Yet the systems responsible for safety and justice routinely fail to recognize and respond to their needs.
Survivors encounter inaccessible services. Defendants move through the legal system without necessary accommodations. People with disabilities are criminalized for behaviors that are misunderstood. And organizations struggle to meet needs they were never adequately equipped to address.
We've spent decades working to change these systems through training, technical assistance, advocacy, and policy reform. That work remains essential.
But what if AI could help us accelerate it?
Imagine a survivor with an intellectual disability who can independently access information about their rights and options in language they understand.
Imagine a small victim service organization able to make its resources accessible in multiple formats without requiring expertise or resources it doesn't have.
Imagine a public defender equipped with tools to better identify disability-related barriers and advocate for necessary accommodations.
And imagine moving beyond helping individuals navigate inaccessible systems to changing the systems themselves.
What if victim service organizations could use AI to examine their intake procedures, eligibility requirements, policies, and communications to identify potential accessibility barriers? Or public defender offices could use similar tools to evaluate whether their practices consistently recognize and accommodate disability?
Rather than waiting until someone encounters a barrier, organizations could identify potential problems, make improvements, and evaluate whether those changes actually expand access.
These applications would require careful testing, meaningful human oversight, and strong safeguards. But they offer an intriguing possibility.
What if AI could help us stop expecting people with disabilities to navigate inaccessible systems and start making those systems accessible to them?
The missing intersection
Conversations about AI are already happening across the criminal justice and victim services fields. Advocates, researchers, and practitioners are grappling with its risks, exploring its potential, and asking important questions about how it should be used.
But I worry that disability—and particularly the intersection of disability, victimization, and criminalization—will be overlooked.
That would be a serious mistake.
People with disabilities and Deaf people are disproportionately harmed by the very systems in which AI is being introduced. Their experiences reveal barriers and inequities that broader conversations about technology, safety, and justice may fail to recognize.
And the risks are significant. Technology that fails to account for disability could misinterpret behavior, reinforce harmful assumptions, compromise confidentiality, or create new barriers to services and justice.
If disability isn't part of the conversation from the beginning, we risk building tomorrow's inequities into today's technology.
We need guardrails, not just innovation.
AI is already being used in systems of safety and justice. Its role will almost certainly grow.
The question isn't whether we embrace or reject it. It's how we ensure that its use expands access, autonomy, and equity rather than reinforcing discrimination, surveillance, and harm.
That requires more than good intentions. It requires clear guardrails, accountability, rigorous evaluation, and meaningful leadership from people with disabilities, Deaf people, survivors, and people impacted by incarceration. It also requires recognizing that some uses of AI may be too harmful to justify.
I've come a long way from my initial irritation with that funder's request. I still believe philanthropy can be too quick to chase the next big thing. But I also believe we have an opportunity—and a responsibility—to help shape how this technology affects the people and communities we serve.
AI can be used to exercise power over people, or it can be used to give people more power over their own lives. We need to help ensure it's the latter.
At Activating Change, we're interested in bringing people together to explore the possibilities, confront the risks, and consider what responsible, disability-centered AI could look like across safety and justice.
Where do you see the greatest opportunities? What concerns you most? And what guardrails do you believe are essential?
Share your thoughts in the comments. Let's start the conversation.