We don’t “wrap” AI. We govern it.
In the rapidly evolving world of Artificial Intelligence, speed is often prioritized over safety. At TheMediator.AI, we take a different approach. We believe that dispute resolution requires more than just intelligence; it requires integrity.
We view ourselves as the “Fifth Party” in your dispute, the architects responsible for the environment in which you negotiate. We take that responsibility seriously.
1. The SafeGate Protocol: A True Fail-Safe
Most AI agents operate “On the Loop”—they generate text instantly, relying on the user to catch errors or hallucinations. This speed creates risk.
TheMediator.AI is built on SafeGate, a proprietary autonomous moderation layer that forces our system “In the Loop”whenever a risk is detected.
- Active Intercept: Every message is analyzed for threats, coercion, intimidation, and self-harm before it is relayed to the other party.
- The Brake Pedal: If SafeGate detects a violation, the mediation stops. The message is blocked, and the sender is required to rewrite it or end the session. We do not allow the AI to “guess” its way through safety risks.
- Coercion Detection: We don’t just look for profanity. We scan for subtle power imbalances and intimidation tactics that silence meaningful negotiation.
Where is the Human Mediator?
In traditional mediation, you have a Third Party (the human mediator) and a Fourth Party (technology like WhatsApp, Zoom, Email, etc;). TheMediator.AI combines these into a single Virtual Facilitator.
This AI provides the structure and impartiality of a human mediator, instantly and at scale. However, we (the developers) remain the Fifth Party.
- Our Role: We design the “SafeGate” architecture that governs the AI.
- The AI’s Role: It acts as an impartial guide, following our strict safety protocols. It does not have its own agency; it operates within the ethical boundaries we have coded.
2. Our Ethical Standards
We align our engineering with the 2022 International Council for Online Dispute Resolution (ICODR) Standards. Here is how we translate those principles into code:
ACCESSIBILITY
- Standard: ODR must be easy to find and use.
- Our Practice: We use ubiquitous platforms (Mobile & SMS) to ensure parties can participate without expensive hardware. We do not assume high technical literacy.
ACCOUNTABILITY
- Standard: The system must be answerable for its outcomes.
- Our Practice: We track Process Integrity, not just “Success Rates.” If a dispute ends because of a safety violation, we log it as a “Protective Pause,” not a failure. We maintain a full audit trail of the logic used to reach an outcome.
COMPETENCE
- Standard: The provider must have relevant expertise.
- Our Practice: Our AI is not a generic chatbot. It is specialized in conflict resolution theory, prioritizing Procedural Justice (the feeling of being heard) over speed. It is tuned to minimize Fabrication. Unlike generic chatbots, our system is instructed to avoid inventing laws. If it cannot verify a specific local regulation, it is designed to ask you for clarification rather than guess.
CONFIDENTIALITY
- Standard: Data must be secure and private.
- Our Practice: Your data is encrypted and compartmentalized. We practice Informational Sovereignty—we do not train our public models on your private dispute details without your explicit consent.
EMPOWERMENT
- Standard: Parties must be able to make informed decisions.
- Our Practice: We use Outcome Modeling. Instead of forcing a settlement, our AI helps you explore your “Best Alternative to a Negotiated Agreement” (BATNA). We empower you to say “No” to a bad deal.
EQUALITY
- Standard: The process must be free of bias.
- Our Practice: We do not strive for “Neutrality” (which can be passive). We strive for Impartiality. Our AI actively monitors for power imbalances to ensure neither side dominates the conversation.
FAIRNESS
- Standard: Due process must be respected.
- Our Practice: We ensure both parties have equal opportunity to speak. Our “Turn-Taking” architecture prevents one party from filibustering the negotiation.
HONESTY
- Standard: Data and intentions must be transparent.
- Our Practice: We use the term “Outcome,” not “Resolution,” because not every dispute ends in agreement. We do not promise legal binding authority where none exists.
TRANSPARENCY
- Standard: The “Fifth Party” (us) must be visible.
- Our Practice: We disclose our methodology. You always know when you are speaking to the AI Facilitator. We clearly mark generated suggestions as “Proposals,” not legal rulings.
3. Methodology & Limitations
To ensure Transparency, we disclose the following limitations of our system:
- We Are Not Lawyers: TheMediator.AI is a facilitator, not a judge or legal counsel. It cannot provide legal advice or issue binding verdicts.
- No “Hallucinations”: We aggressively tune our models to avoid fabricating facts. If the AI seems unsure about a specific local law (e.g., tenancy rules in a specific zip code), it is instructed to pause and ask you for clarification rather than guess.
- Human Review: While the mediation is automated, you always have the option to End Mediation and receive a transcript summary to take to a human lawyer or mediator.
4. ODR Standards Self-Certification
Statement of Self-Certification
“We self-certify compliance with the Online Dispute Resolution (ODR) Standards and will accept inquiries and respond with transparency about how we comply with each of the ODR Standards.”
— TheMediator.AI Team
We are proud to stand with the global community of ODR practitioners in building a safer, fairer digital justice system.
Need to Report a Safety Concern?
If you believe a mediation session violates these standards or poses a safety risk, please contact us immediately.
Important Disclaimer: Probabilistic, Not Deterministic
TheMediator.AI uses Large Language Models (LLMs) to facilitate communication. While we aggressively tune for accuracy, AI is probabilistic, meaning it generates responses based on patterns and data.
- Verification Required: The AI may occasionally generate incorrect information (“hallucinations”). Users are responsible for verifying all settlement terms, dates, and local laws before agreeing to an outcome.
- No Legal Advice: The Virtual Facilitator provides process guidance, not legal counsel. Suggestions made by the AI are for negotiation purposes only.
- Limitation of Liability: By using this service, you acknowledge that TheMediator.AI provides the platform for negotiation (Process Integrity) but is not liable for the substantive terms of any agreement reached.