A new artificial intelligence system is being developed to assist in arbitration, with the potential to streamline legal disputes and make them more accessible. The American Arbitration Association, led by former Michigan Supreme Court Chief Justice Bridget McCormack, has introduced an AI Arbitrator that utilizes OpenAI’s models to draft decisions based on document-based disputes. This tool aims to reduce the burden on human judges while ensuring a quicker resolution process, potentially cutting the time taken from 60-75 days to just 30-45 days for certain cases. However, concerns persist regarding the accuracy and potential biases of AI in legal settings, with experts emphasizing the importance of human oversight in the judicial process.
Why do we care?
Nobody is saying the AI replaces arbitrators. But once the machine writes the first version, the human role quietly changes. You’re no longer deciding—you’re correcting.
That’s fine at small scale. It’s dangerous at volume.
Translate this to IT services. If you deploy AI to draft security recommendations, compliance findings, or incident assessments, and your team “reviews” them under time pressure, you’ve already crossed the line. You own the outcome, but you didn’t originate the logic. The only safe version of this model is one where MSPs control the prompts, models, escalation thresholds, and audit trails.
That’s how MSPs end up holding liability for decisions they didn’t fully control.
Arbitration is just the early, polite version of the problem. The same pattern is coming to insurance claims, HR disputes, vendor audits, and security triage.
If MSPs copy this model without governance—treating AI drafts as efficiency tools instead of authority shifts—they’ll absorb risk they never priced, defend decisions they didn’t make, and lose client trust when something goes wrong.
Faster isn’t safer. Automation isn’t neutral. And “human oversight” isn’t a shield if you don’t control the system underneath it.