The practice exists. A short name for it doesn't.
Operators already run a model from a rival lab against their AI's work. OpenAI ships an official plugin that drops Codex into Claude Code as a reviewer. Practitioners who write about the pattern reach for their own multi-word phrases, each a little different. You can't ask a team to "do the thing where the other lab's model argues," and a practice without a shared short name is hard to put in a checklist, budget for, or audit.
So, a name:
Cross-Lab Adversarial Review (CLAR) is the practice of putting a model from a different lab against the work — an AI's, or a person's — as an adversarial reviewer. The rival needn't be smarter; it's differently trained, so its blind spots are less likely to match. That is the value — decorrelation beats intelligence: their agreement carries real information, and the disagreement they can't resolve marks the call that still needs a human.
Why "Cross-Lab" Is the Load-Bearing Word
Not "adversarial" — telling any reviewer to refute rather than review is good practice with any model. The part CLAR is betting on is the different lab — model provenance — and that bet has research behind it.
The premise — that shared blind spots are the risk, and that crossing labs reduces them — is grounded in two findings. Great Models Think Alike and this Undermines AI Oversight (arXiv:2502.04313) finds model errors are substantially shared, and more so among more-capable models — the reason self- and same-model review is a weaker control than it looks. Correlated Errors in Large Language Models (Kim et al., arXiv:2506.07962) measures the overlap directly: models agree on the same wrong answers well above chance, with the correlation lower — but far from zero — across labs. Neither paper tests an adversarial-review protocol; they establish the problem CLAR responds to, not its effectiveness. Cross-lab review buys a discount on shared blind spots, not an exemption — which is why agreement stays evidence, not proof.
The first paper puts a direction on it: model errors, in its own framing, are becoming more similar as models grow more capable. Which turns the cheapest check you have — a model reviewing itself, or its sibling — into a control that's quietly losing value: weakest exactly where it looks free.
A limitation of the term: "cross-lab" names the intent — a reviewer trained by a different lab, which is where the decorrelation comes from — but a tool selects by API provider as the practical proxy for lab. The two usually coincide, but not always (Azure serves OpenAI's models), so the provider setting is the means and the different lab is the point.
Three Rules Make It a Discipline
The definition stops short of any tool on purpose; the practice should outlive whatever implements it. But without discipline, a cross-lab review decays into two models trading opinions — confident, articulate, unfalsifiable. My implementation, below, is opinionated about three rules.
Every finding carries anchored evidence. A location and an observation — a line number, a named standard, a reproduction. Never a vibe. Evidence is what lets the first model check the claim against the actual artifact instead of debating tone.
A rejection needs contradicting evidence too. Verification is symmetric. The first model saying "I disagree" doesn't dismiss a finding; if it can't produce proof against it, the finding survives — recorded as a deadlock rather than waved away. Agreeableness corrupts a review in both directions.
The escalation is the deliverable. Most findings resolve — verified and fixed, or refuted with proof. What reaches a person is the short list neither model could settle. If everything escalates, you've built a forwarder. If nothing escalates, you've built a rubber stamp.
Concretely: while building this site, Claude and Codex deadlocked on a diagram color — one citing a contrast standard, the other the site's design system, both right about their own half. Neither could win on the evidence, so the question reached me: the one decision in that batch of work that was actually mine. The dozen clean catches around it never did.
What CLAR Is Not
A new discovery. The human antecedent is adversarial collaboration — opponents designing a fair test together, with an arbiter — as practiced by Mellers, Hertwig, and Kahneman (2001); CLAR points the same posture at models. The AI lineage approaches it from other angles: AI safety via debate (Irving et al., 2018) has competing agents argue before a human judge; multiagent debate (Du et al., 2023) improves answers by argument among instances of one model; LLM-as-a-judge (Zheng et al., 2023) has an external model score outputs; and PoLL (Verga et al., 2024) already juries across providers, aimed at judge bias and cost. Practitioners have described the cross-vendor adversarial version under longer names — "cross-model adversarial review" (Vaughan), "adversarial cross-model review" (Wade). CLAR is a short label and a discipline for a practice that was already emerging. The term was introduced here, July 2026.
The Reference Implementation
Impasse is the reference implementation of CLAR — open source, MIT-licensed, and deliberately dull in its mechanics. It hands the artifact to a rival-lab model with instructions to refute, requires structured findings with anchored evidence, and routes only genuine deadlocks to a person. It reviews more than code: a decision memo, an essay, a research claim, a dataset. And it is read-only on the work — the reviewer criticizes; it never edits.
The field story — what a rival reviewer caught over a day of real work, what it got wrong, what the arguing looked like — is in AI's Second Opinion.
What It Costs
Two bills, for a start, and a real review loop burns tokens. Wiring one agent to call another reliably takes work. And the failure mode is quiet: let the two models share a prompt, a style guide, a set of assumptions, and you've collapsed them into one reviewer with two logos — you've paid to agree with yourself more confidently. Decorrelation is something you protect, not something you get free with the second login. A CLAR pass is only as good as the difference between the two models you feed it.
AI Workshop for CEOs
Deciding which of your team's decisions deserve a CLAR pass — and which disagreements should reach a human — is exactly the kind of operating discipline the workshop builds with your team. Three hours live with a group of 8 CEOs, plus a 1-on-1 to fit it to how your work actually gets done.
Reserve Your Seat →The day-by-day account of this practice in the field — the deadlocks, the catches, the costs — is in AI's Second Opinion, and a single catch worth its own write-up — a spam filter that passed its own tests but would have dropped real leads — is in Green Tests, Two Bugs. Why the harness around a model matters more than the model is in the harness beats the model; how operators structure whole agent fleets is in your AI agents need an org chart.