OpenAI and Sullivan & Cromwell co-develop Agreement Analyzer for contract review

OpenAI said in a post on its website that it worked with the law firm Sullivan & Cromwell to develop Agreement Analyzer, an AI tool designed to identify issues in a draft agreement.

As OpenAI describes it, the tool targets a specific step in the contract workflow: catching problems and open questions before an agreement is finalised, rather than after its terms take effect.

The hard part of that job is judgement. What counts as an issue, which problems matter most and which risks are acceptable depend on the deal type and the firm's own experience, a body of knowledge that general-purpose rules struggle to capture.

The more interesting signal is the form of the collaboration. The tool was built by the firm that will use it together with the model provider, which suggests the operating standards came from day-to-day legal work rather than from a model's default behaviour.

For OpenAI, legal work is one of the most document-heavy corners of professional services. Contracts, regulatory filings and deal memoranda are long, structured and expensive to get wrong, which is why they have been a shared testing ground for model vendors and enterprise clients.

For law firms, adopting a review tool is not only a question of speed. Once such a tool enters the workflow, liability questions follow: whether flagged issues are acted on, and who answers for the ones that are missed.

OpenAI's disclosure so far is limited to the tool itself. It has not detailed how Agreement Analyzer is delivered, which kinds of agreements it covers, or whether it will be available beyond Sullivan & Cromwell.

The question to watch is whether customised, firm-built tooling of this kind can be replicated elsewhere. If it can, AI-assisted contract review moves from a one-off experiment toward a standard part of legal practice; if not, it looks more like a bespoke internal project.