Courts keep sanctioning lawyers over citations nobody checked.
The model can be measured first.
We are an independent testing lab. We measure where the language model your firm hosts fabricates, using the research patterns your attorneys actually run, so verification effort goes where the risk is.
Opinion 512 named the duties. The sanctions lists keep growing.
The pattern in the sanctions orders is consistent: confident output, plausible citations, nobody checked. ABA Formal Opinion 512 put the verification duty in writing, and a firm running its own model for confidentiality reasons is left with a practical question. Verified against what? No firm can check everything, but it can find out where its model fabricates and check hardest there.
What the instrument does is read the model's internal signals while it answers the research patterns your attorneys actually use, on a model the firm hosts or designates. It maps where the model fabricates and how often, and every flagged answer arrives in full so the firm can verify it directly. The deliverable is a signed, dated screening measurement. Detection performance so far: 0.852 AUC on models the instrument had never seen, fabricated-entity discrimination, length-controlled, leave-one-model-out. Method paper DOI: 10.5281/zenodo.21365654.
No client files. No privileged material. Five business days.
It never touches client files or privileged material. It needs representative question types, research patterns the way your attorneys actually phrase them, and a model endpoint the firm designates: a self-hosted model or a vLLM-class private-cloud deployment the firm controls. Private Azure OpenAI exposes probabilities on output tokens only, so that takes behavioral testing instead. Consumer AI tools the firm doesn't control are out of scope, and we say so.
The engagement is written-first and async. Every question gets answered in writing, on the record.
The price is published and it never depends on what I find.
The LLM Validation Report is $9,500. Continuous Assurance is $6,500 per quarter for one model. The pre-deployment Model Selection Study is $4,500 for two candidates, up to $7,500 for four. Full terms are on the pricing page. No work begins without a signed engagement letter, and no fee is ever success-based.
Questions firms ask first
Does the test see client files or privileged material?
How does this relate to ABA Formal Opinion 512?
Does self-hosting change the test?
Can the results help after a citation incident?
Read the report before you decide anything.
Email us and the sample validation report comes back the same day. It is the same signed, dated document your risk committee would review.
Email the lab