Just as eDiscovery created chain of custody for digital evidence, Anchor BlackBox™ creates Chain of AI Custody™ for AI-assisted work product. Six chains. One unbroken record. Court-producible on demand.
Every AI-assisted matter in your firm generates six independent chains of evidence. AIDR™ and AICR™ record each chain automatically. At session close, the six chains are bound into a single tamper-evident ledger, the Chain of AI Custody™ record.
Which model, which version, which API endpoint, which inference parameters were used on each request. No ambiguity about what answered your question.
Which attorney made which request, applied which correction, and provided which approval at each step. The attorney of record is always identified.
Every version of every document the AI touched: input, intermediate, and final, with diff-level provenance. You can account for every word.
How and when each AI-assisted work product was delivered to the client, opposing counsel, co-counsel, an expert, or the court, with delivery confirmation and receipt log. Enforced on the way out by Outbound Sentinel™ ↓.
Every governance checkpoint: competence acknowledgment, supervision sign-off, client authorization event, final release. The audit trail courts demand.
A cryptographic timestamp on every event. The full chronology sealed at session close. Tamper-evident and court-producible from the moment the session ends.
A document doesn't need to be new to be dangerous. It needs to still contain something Anchor BlackBox™ flagged, at any point in the matter's life, that nobody caught this time.
Court e-filing upload. Email to opposing counsel, co-counsel, or an expert witness. Anything leaving the firm's privileged boundary triggers Outbound Sentinel™ on the final version, at the moment the last reviewer sends or uploads it.
Outbound Sentinel™ does not search for exact text. It runs a deep contextual comparison of the outgoing document against every non-Green flag ever raised on that client-matter, catching the same substance even if it was reworded, restructured, or paraphrased since it was first flagged.
It does not matter that the paralegal, the associate, or the supervising attorney already reviewed and cleared that content earlier in the matter. Outbound Sentinel™ re-evaluates from zero, every time, at the moment of delivery.
A match routes an alert, with full risk detail, to the firm's top-level partner review stakeholders, not only the matter's original team. Only a document that clears Outbound Sentinel™ earns the Green Chain Score™ and is released.
The Delivery Chain is where every other chain gets tested. AIDR™ and AICR™ can record a flawless session, and a firm can still ship a document that quietly carries something flagged months earlier, on a different draft, by a different person who has since moved on. Outbound Sentinel™ is what keeps that from being an assumption instead of a certainty. See how the flag first gets raised, at the moment the AI responds, on AICR™ →.
When eDiscovery emerged in the early 2000s, firms that had no process for preserving and producing digital evidence faced sanctions, adverse inference instructions, and malpractice exposure. The firms that built the infrastructure first were protected. Chain of AI Custody™ is that moment, for AI.
Courts and clients no longer ask whether you used AI. They ask for the record. Chain of AI Custody™ is that record.
An AI governance policy tells you what your firm intends. Chain of AI Custody™ proves what your firm actually did, matter by matter, session by session.
A work product with no Chain of AI Custody™ is an Orphan Work Product. In a discovery challenge, its absence is itself evidence of negligence.
Every matter. Every session. Automatic. Request a demo and see all six chains built on a live AI session.
Request Access