Research 4 min read

Pathological dispute clauses: what fails, and how we flag it

A look at the defective arbitration and jurisdiction clauses we see across our index, and how Review detects institutional mismatches before signing.

2026 · 10 · 09·admin

Dispute resolution provisions are often drafted last. Negotiators spend months on economic terms, earn-outs, and warranties, leaving the governing law and jurisdiction clauses to the final hours of a transaction. The result is the midnight clause: a rushed, often copy-pasted paragraph that dictates how and where disputes will be resolved.

When these clauses are defective, they are termed pathological. A pathological clause contains ambiguities or contradictions severe enough to derail the dispute resolution process before it begins. Parties are forced to spend months, and considerable legal fees, arguing over where they are supposed to argue.

Across the 2.4 million contracts now processed by our systems, dispute clauses are a persistent source of unforced errors. In a recent analysis of 100,000 executed commercial agreements in our 2026 matter index, we found that 3.4% contained a material defect in the dispute resolution tier.

Here is what fails, and how we train ZAAN-7B Counsel to catch it.

Non-existent institutions and legacy ghosts

The most common pathology we observe is the phantom institution. A drafter pulls a clause from a fifteen-year-old precedent and fails to update the arbitral body.

We routinely see contracts directing disputes to the “International Chamber of Commerce of Geneva” (a conflation of the Paris-based ICC and a Swiss seat) or the “Swiss Chambers’ Arbitration Institution” (which was restructured and renamed the Swiss Arbitration Centre in 2021). We also see references to regional arbitration centres that have been dissolved or absorbed by other entities.

Keyword matching is insufficient here. A standard rules-based system sees “Arbitration” and “Geneva” and passes the clause. ZAAN-7B Counsel reads the entity name against a maintained index of active arbitral institutions. If a draft names a defunct or non-existent body, Review 4.2 flags it as a High severity issue. The inline redline replaces the phantom institution with the default arbitral body specified in the firm’s playbook, adjusting the surrounding boilerplate to match that institution’s recommended model clause.

Conflicting tiers and optionality

Multi-tier dispute resolution clauses—where parties must negotiate, then mediate, then arbitrate—are standard practice. They become pathological when the transition triggers are poorly defined.

A clause stating that parties “shall negotiate in good faith for a reasonable period before commencing arbitration” is an invitation to preliminary litigation. What constitutes a reasonable period? When does the clock start?

Review flags transition triggers that lack hard integers. If the model detects a mediation or negotiation tier without a strict time limit (e.g., “30 days from written notice of the dispute”), it generates a Note severity flag, prompting the reviewer to anchor the timeline.

More severe is the problem of conflicting optionality. We occasionally see clauses stating that disputes “shall be resolved by arbitration in London,” followed two paragraphs later by a severability clause stating that “the courts of England and Wales shall have exclusive jurisdiction over any dispute.” This usually happens when a drafter pastes an arbitration clause into a master services agreement but forgets to delete the legacy jurisdiction boilerplate. Review identifies the contradiction and forces the reviewer to choose one path, proposing the deletion of the other.

Asymmetry and governing law

Asymmetric jurisdiction clauses grant one party the right to sue in any competent court, while restricting the other party to a single venue or to arbitration. These are common in finance documents, allowing the lender maximum enforcement flexibility.

The pathology arises not from the clause itself, but from its interaction with the governing law. French courts, for example, have historically taken a restrictive view of asymmetric clauses (the Rothschild doctrine), often rendering them unenforceable. English courts generally uphold them.

Because ZAAN-7B Counsel is trained for clause-level reasoning, it does not evaluate the dispute clause in isolation. It cross-references the jurisdiction provision with the governing-law provision. If the contract is governed by French law and contains an asymmetric jurisdiction clause, Review raises a High severity flag, warning the drafter of the enforceability risk.

Downstream effects in Simulation

The clarity of a dispute clause directly impacts our Trial & Arbitration Simulation product. To instantiate a mock tribunal, Simulation requires a defined venue, seat, and governing law. These parameters dictate the procedural rules the AI chair and opposing counsel will follow.

When a legal team uploads a contract with a pathological clause into Simulation, the system cannot silently infer the venue. Instead, the AI tribunal raises the defect as a preliminary jurisdictional objection. Before the team can test their substantive arguments on liability or damages, they must first survive a simulated hearing on whether the tribunal even has the authority to hear the case.

This mirrors reality. A broken clause guarantees a jurisdictional fight.

Catching these defects during the review phase is a matter of basic hygiene. A firm’s playbook should not just dictate what the dispute clause says, but ensure it actually functions when invoked.

A contract is only as binding as the mechanism used to enforce it.

See it on a contract you have already reviewed.

Send us a draft your team has already redlined and we will show you what ZAAN catches, and what it misses.