AI DISPUTE MEDIATION
Understanding the technology to understand the dispute
What is AI Dispute Mediation?
These are disputes that arise from artificial intelligence, not disputes settled by it. A model was trained on proprietary material without permission. A vendor’s system did not do what the contract said it would. An automated screening tool rejected a class of applicants. Software made a decision, and someone was harmed by it.
The technology is new. The underlying disputes are not. A training data claim is a copyright case. A failed AI deployment is a contract case. An algorithmic hiring claim is an employment case. What changes is that the evidence sits inside a system that is difficult to examine, and that neither side may be able to explain exactly what it did or why.
I mediate these matters throughout Florida and Texas, remotely or in person.
-
Training data and IP claims, including whether protected material was used to build a model and on what terms
-
AI vendor and licensing disputes over performance, scope, indemnity, and ownership of outputs
-
Employment claims involving algorithmic decision-making in screening, hiring, promotion, and termination
-
Liability when an automated system fails and someone is harmed by the result
-
Professional and business liability arising from the use of AI tools, including work product produced without adequate review
-
Trade secret and confidentiality claims when sensitive material is disclosed to an AI system that retains or reuses it
Why Choose Mediation for AI Disputes?
These matters are expensive to litigate and slow to resolve, and much of that cost is spent establishing facts that may still be contested at the end. Mediation lets the parties settle the dispute more quickly and less expensively.
Confidential
Proprietary models, training data, and system architecture stay out of the public record. For many parties this is the single most important consideration, and it is one a trial cannot offer.
Cost-Effective
The technical discovery in these matters is the most expensive part of the case. Mediation can resolve the dispute before that expense is incurred.
Retain Control
When neither side can be confident how a court will rule, a resolution both sides helped design is usually worth more than a coin flip with a two-year wait.
Flexible
Courts only award damages. Parties in mediation can agree to licensing terms, model retraining, attribution, audit rights, or indemnity — remedies no judgment can produce.
Why These Disputes Are Harder Than They Look
Three things make an AI matter difficult in ways an ordinary technology dispute is not, and each of them affects how the case should be mediated.
Nobody can fully explain the system. With conventional software, someone wrote the code and can tell you what it does. Large models train themselves, so even the people who built it only have a vague idea of how it works. Once it is running, no one can state precisely how a model reached a particular output, and you cannot get a reliable answer by asking it. That means causation is often genuinely contested rather than merely disputed, and both parties may be arguing from inference.
The evidence is expensive and slow to develop. Establishing what a model was trained on, how it was tuned, or why it produced a specific result can require discovery that costs more than the claim is worth, and may run into trade secret objections on the other side. Parties frequently end up litigating access to the evidence long before they litigate the merits.
The law is unsettled. These cases are being decided now, and counsel on both sides are working without the settled authority they would have in almost any other matter. That uncertainty cuts both ways, which is precisely the condition under which a negotiated resolution tends to serve both parties better than a verdict.
Brandon’s Experience with Artificial Intelligence
Brandon holds the Google Data Analytics Certificate and the Google Generative AI Leader Specialization, together with IBM’s Generative AI Essentials and Generative AI: Prompt Engineering through Coursera. For the employment side of these disputes he has also completed Employment Contracts through the University of Pennsylvania Carey Law School and Human Resource Management and Leadership through Macquarie University.
He teaches “Winning the Mediation Game,” a continuing education program approved for CLE credit in Florida and Texas, which includes the use and limits of AI in dispute resolution — where these systems fail, why their reasoning cannot be audited the way conventional software can, and what that means for anyone relying on their output.
Disputes arising from AI are an extension of work Brandon already does in technology, intellectual property, and employment mediation rather than a separate specialty. What he brings to them is the ability to follow the technical argument rather than take a summary on faith — the same thing that lets him read a medical record in a personal injury case instead of relying on someone else’s characterization of it.
For related matters, see technology and intellectual property mediation and workplace mediation.
Brandon is a Florida Supreme Court Certified Circuit and County Court mediator, also trained in Texas.
Frequently Asked Questions
Do you mediate disputes about AI, or do you use AI to mediate?
Disputes about AI. I mediate matters in which an artificial intelligence system is the subject of the dispute — what it was trained on, what it produced, or what it cost someone. I do not use an AI system to conduct mediations or to decide anything, and I do not put confidential case material into one. The reasoning behind that is part of what I teach in Winning the Mediation Game.
How do you evaluate a case when neither side can explain what the system actually did?
That is the defining feature of these matters. With conventional software, someone wrote the code and can testify to it. With a large model, no one can state precisely how a particular output was reached — not even the people who built it. I treat causation as genuinely open rather than as a fact one side is concealing, which moves the conversation from proving fault to pricing uncertainty. Pricing uncertainty is something both sides can actually do in a mediation, but a jury cannot.
What happens when the discovery needed to prove the claim costs more than the claim?
That is common, and it is usually the strongest argument for mediating early. Establishing what a model was trained on, how it was tuned, or why it produced a particular result can require expert work that outruns the value of the case, and it often meets trade secret objections that create a second fight before the first one begins. Parties who see that coming can resolve the dispute on terms they design, rather than spending the value of the case discovering facts that may still be contested at the end.
See our complete FAQ for more.
Schedule a Mediation
Don’t spend years in court when you could have a settlement in hours.
