Legal and compliance documents for AI training
Legal and compliance data covers how companies negotiate, approve and document obligations: contract drafts and redlines, signed agreements with clause data, and completed security questionnaires. AI labs license it to train legal and compliance agents on real negotiations rather than final public filings.
Listings
| Data | Typical sources | Modality | Availability |
|---|---|---|---|
| Contract drafts, redlines and negotiation threads | Ironclad, DocuSign CLM, Juro | Documents and text | On request |
| Executed commercial contracts with clause data | Ironclad, DocuSign CLM, Agiloft | Documents and structured data | On request |
| Security and compliance questionnaires | Vanta, Drata, OneTrust | Documents and structured data | On request |
What’s included
- Successive contract drafts with tracked changes and comments
- Executed agreements with extracted clauses, dates and obligations
- Completed security and compliance questionnaires with answers and follow-ups
Public datasets and licensed data
Public contract datasets are built from final agreements, mostly SEC filings or documents posted online. They contain no drafts, redlines or negotiation history, and no internal approvals. Licensed data adds how terms actually moved during negotiation.
| Public dataset | Released | What it contains | Limits for enterprise use | License |
|---|---|---|---|---|
| CUAD (Contract Understanding Atticus Dataset) | The Atticus Project, 2021 | 510 commercial contracts from SEC EDGAR with 13,000+ expert labels across 41 clause types. | Final agreements filed publicly; no drafts, redlines or negotiation history. | CC BY 4.0 |
| LEDGAR | LREC 2020 | Contract provisions from 60,540 agreements filed on EDGAR in 2016–2019. | Publicly filed clauses only. | See source |
| ContractNLI | Stanford and Hitachi America, 2021 | 607 non-disclosure agreements, each checked against 17 fixed statements. | One agreement type and a fixed set of questions. | CC BY 4.0 |
Preparation and rights
Preparation requirements are agreed with each supplier before any work begins. For this kind of data they typically include:
- Redacting, excluding or clearing with the counterparty any agreement whose confidentiality terms restrict disclosure
- Excluding privileged legal advice
- Redacting counterparty names and identifying deal terms
Every dataset has a documented owner and confirmed licensing rights, and is delivered only after an executed agreement and the supplier’s authorization. See data governance on sourcex.si.
Use cases
More on sourcex.si
Questions
Doesn’t contract confidentiality rule this out?
It’s the main check, not a blanket barrier. Agreements whose confidentiality terms prohibit disclosure are excluded or cleared with the counterparty, and the rest are redacted before delivery.
Can law firms license client documents?
Only with the client’s informed consent, because lawyers owe clients confidentiality. This catalog focuses on companies’ own in-house contracting.
Need legal and compliance documents?
Describe what you need: domain, volume, history, format and licensing terms. SourceX looks for companies that hold it and can license it.