AI Can Write a Legal Motion, But Can’t Guarantee a Father Gets to See His Child

By Kenneth Braswell, CEO, Fathers Incorporated

In our work with fathers, we often encounter fathers holding court papers they don’t understand. Sometimes the papers concern child support. Sometimes custody. Sometimes parenting time or legitimation. 

The details change, but the look on the father’s face is familiar. He knows the document matters and that his relationship with his child may depend on what happens next. What he doesn’t always know is what some of the words mean, what he is supposed to file, what to say when his name is called, or whether he can afford someone who can help him.

For years, the gap between having a legal problem and being able to navigate it has been one of the quiet realities of responsible fatherhood work.

Artificial intelligence is beginning to change that.

Fathers Are Turning to AI for Help With Court Hearings

Today, a father with a smartphone can ask an AI system to explain a child support order, translate legal language into plain English, organize a timeline, identify questions to ask an attorney, or produce something that looks remarkably like a legal motion. What once required hours of research, a knowledgeable advocate, or hundreds or thousands of dollars in legal assistance can sometimes appear on a screen in seconds.

That can sound like access to justice, but a fascinating new paper suggests we should be more careful about what we call access.

In “Artificial Access to Justice: AI and the Surge in Pro Se Litigation,” Dr. Or Cohen-Sasson examines roughly 2.8 million federal civil filings from 2008 through 2025 and nearly 13,000 complaints filed by people representing themselves. For about 15 years, self-represented plaintiffs accounted for roughly 11% of federal civil filings. After generative AI became widely available, that number began climbing, reaching 22.3% by the final quarter of fiscal year 2025.

The study also found that complaints showing signs consistent with AI assistance looked more professional than traditional self-represented filings. They used more citations and carried more of the polish we associate with lawyers. However, these complaints did not perform better. They were dismissed more frequently, terminated earlier, and were no more likely to prevail than their non-AI-assisted counterparts. The study’s author calls this the “litigation-efficacy paradox” — more entry, more legal form, but no greater access to remedy.

The study concerns federal civil litigation. It does not study family courts, child support cases, custody disputes, parenting-time proceedings, or fathers specifically, but after more than two decades working with fathers, I cannot read its findings without thinking about the men I’ve watched struggling to navigate those systems.

Fathers Incorporated’s own experience tells us how central legal systems can become to the lives of nonresident fathers. Among fathers participating in our Gentle Warriors Academy from 2021 through 2023, the leading service needs included legitimation, child support, parenting time, and broader family legal matters. Nearly 450 participants expressed a need for legitimation services during that period.

These numbers represent more than legal problems. Behind them are fathers trying to answer very human questions:

• Can I see my child?
• Can I participate in decisions about school?
• What happens if I cannot afford the amount of child support that has been ordered?
• How do I ask the court for parenting time?
• What rights do I actually have?

In Georgia, those questions can become especially complicated for unmarried fathers. Establishing paternity and establishing legal rights have historically operated as separate matters. A father may have financial responsibilities toward his child while still needing to pursue legitimation before he can obtain the legal standing associated with custody and parenting time. Our research has repeatedly found that fathers can find this distinction confusing, expensive, and intimidating.

Now place generative AI in the middle of that experience.

The Downsides to Fathers’ Use of Generative AI in Court Proceedings

A father who can’t afford an attorney today may be able to generate a motion tomorrow. That’s progress of one kind, but it’s not necessarily justice.

Family court isn’t simply a paperwork system. It’s a human system in which facts, credibility, procedure, evidence, relationships, safety, history, and judicial discretion can all matter. A document can sound legally sophisticated while the person whose name appears at the bottom doesn’t understand what it says.

Dr. Cohen-Sasson describes part of this problem as “candor without comprehension.” Our legal system generally assumes that the person signing a filing has read it, understands its assertions, and stands behind its factual and legal claims. Generative AI complicates these assumptions because the person who has experienced the events, the person signing the document, and the entity choosing the words may no longer be the same.

Imagine what this could mean for a father.

He walks into court carrying a beautifully formatted motion. It cites statutes and cases. It uses phrases he has never used in conversation. It sounds like a lawyer wrote it. Then the judge asks him a few questions:

• “What evidence supports this allegation?”
• “What exactly are you asking this court to order?”
• “Why does the authority you cited apply to your case?”

The document may be fluent, but the father may still be lost.

This isn’t an argument against AI. It’s an argument against confusing the ability to produce legal language with the ability to navigate a legal system.

The paper identifies another phenomenon that should concern anyone thinking about AI and self-represented parents. Cohen-Sasson calls it the “polish penalty.” Courts have historically afforded some latitude to self-represented litigants precisely because they are not trained lawyers. Their imperfect language can signal that lack of training. But when AI makes an unrepresented person’s filing look professionally drafted, the appearance of expertise may obscure the reality that the person standing before the judge remains untrained. The paper argues that a protection designed for the legally inexperienced works differently when technology makes the inexperienced look experienced.

Policy Implications of Generative AI as Legal Counsel

We don’t yet know whether this phenomenon is occurring in family courts or whether fathers experience it differently from mothers. That research needs to be done, but we should be asking the question now.

The responsible fatherhood field has spent years telling policymakers that access and engagement are not the same thing. A father can live outside his child’s household and remain deeply involved. A father can live inside the household and be emotionally disengaged. Marriage is a family structure. Presence is a behavior. Engagement is a relationship.

Perhaps we need a similar distinction in our conversation about technology and justice:

• Access to information is not access to representation.
• Access to document creation is not access to legal understanding.
• Access to the courthouse is not access to a remedy.
• Access to a motion is not access to a child.

There is an additional danger here for low-income fathers. Many already enter legal systems carrying disadvantages that existed long before artificial intelligence arrived. A major state-by-state examination of policies affecting fathers found that low-income nonresident fathers often face overlapping challenges involving employment, housing, education, incarceration histories, child support obligations, and limited contact with their children. The same report described how unmarried fathers can receive child support orders without simultaneously receiving clear parenting-time arrangements.

AI didn’t create these problems, but it may expose them, and that may be the most valuable lesson from Cohen-Sasson’s research. The paper resists treating artificial intelligence itself as the underlying problem. Instead, it argues that AI is revealing weaknesses in an access-to-justice system that already fails many people who can’t afford legal help. 

Simply prohibiting AI-assisted filings would remove this new form of assistance without repairing the institutional failures that make people desperate for that assistance in the first place. I agree, but we need to go further when children and families are involved.

A Warning for Dads: Avoid Using AI for Legal Representation

The question isn’t whether fathers should use AI. (They already do and will continue.) The better question is what we want AI to help fathers do.

I see enormous potential when AI functions as a translator: 

• A father can take complicated legal language and ask for a plain-language explanation. 
• He can ask for definitions of key terms like “arrears,” “modification,” “service,” “jurisdiction,” or “parenting plan.” 
• He can organize questions before meeting with an attorney or case manager.

AI can also become a navigator. It can help a father organize documents, construct a timeline of events, prepare questions, identify information he may be missing, and better understand the process he is entering.

Where I become more cautious is when navigation quietly becomes representation.

An AI system should not leave a father believing that because it generated a persuasive motion, he now possesses the judgment of an experienced family-law attorney. Legal representation is not simply the production of words. A good lawyer sometimes tells a client something an algorithm may be reluctant to say: 

• Do not file this. 
• You do not have enough evidence. 
• Your expectations are unrealistic. 
• You’re misunderstanding the order. 
• Your behavior is contributing to this conflict. 
• There’s a safer and less adversarial way to resolve this.

Sometimes access to justice requires assistance. Sometimes it requires judgment. And sometimes it requires someone willing to tell us no.

Strategic Use of AI in Fatherhood Programs

This suggests a different future for responsible fatherhood organizations. We shouldn’t become law firms, nor should we encourage fathers to treat artificial intelligence as one. We can, however, become better bridges between fathers, technology, courts, child support agencies, mediators, legal-aid organizations, and qualified attorneys:

• Imagine a fatherhood program in which a dad can bring in an order he doesn’t understand and receive help developing the questions he needs to ask. 
• Imagine AI-assisted legal education combined with human case management. 
• Imagine court forms written in genuinely understandable language, guided interviews that help fathers organize relevant facts, and referral systems that recognize when a question requires an attorney. 
• Most importantly, imagine requiring comprehension before submission.

Before an AI-assisted document leaves a fatherhood program, the father should be able to explain in his own words what it says, what facts it asserts, what relief it requests, and what may happen next. That’s not simply legal literacy; it’s dignity.

There’s also an opportunity here for researchers. We need to know how fathers are already using generative AI in family-law matters: 

• Are they using it to understand child support orders? 
• Are they generating custody or parenting-time motions? 
• Does AI improve their comprehension? 
• Does it increase filings? 
• Does it reduce mistakes? 
• Does it produce false confidence? 
• Are AI-assisted fathers more prepared when they appear before judges, or merely carrying better-looking paperwork?

And perhaps the most consequential question is this: Does AI reduce the information gap while leaving the representation gap untouched?

Data can tell the truth without telling enough of the truth. A future study might show that thousands more fathers are filing motions with AI assistance. That number alone would tell us very little about whether children are experiencing stronger relationships with safe, responsible, nurturing fathers.

The purpose of better technology should be better support. The purpose of better data should be better support. And the purpose of family court should never become producing more litigants who know how to file papers. It should be helping families reach lawful, safe, child-centered resolutions.

There’s no algorithm that can determine whether a father will keep his promises after leaving the courthouse. No chatbot can manufacture trust between co-parents. No language model can love a child, repair years of conflict, make a support payment, attend a parent-teacher conference, or show up Saturday morning when a child is waiting at the window.

AI may help a father find the words, but human beings still have to build the relationship. And if this technology is going to transform access to family justice, our measure of success can’t simply be whether more fathers learn how to file motions. We have to ask what happens after the motion is filed:

• Does he understand what he signed?
• Was the process fair?
• Is the child safe?
• Can the parents communicate better?
• Did the system help him become more responsible rather than simply more litigious?

And, when healthy father engagement is in that child’s best interest, did all of this technology ultimately help a father do the one thing the motion was written for in the first place?

Be a father to his child.