Validate · Assess & Decide · Writing & Content
GenAI IP risk contract checklist validator
A JSON clause register covering all five canonical GenAI IP risk categories passes validation: every category is present exactly once, each has a clause excerpt of minimum length, a contract reference, and a status from the allowed enum; no category is misspelled or duplicated.
You receive: A JSON object with 'contract_name' (string), 'counterparty' (string), 'review_date' (ISO date), and 'ip_clauses' (array of 5+ objects). Each clause object: category (enum: training-data-license|indemnification|confidentiality-prompt|output-ownership|ai-use-disclosure), clause_excerpt (string >=50 chars), contract_section_ref (string, e.g., '§7.2'), status (enum: present|absent|under-negotiation).
Part of Protect IP & Data
Example
A sample of what this play produces. Your result is generated for your inputs.
Contract name
Northlake Model Labs Master Inference Services Agreement (draft v3, pre-signature review)
Counterparty
Northlake Model Labs, Inc. (fictional example vendor)
Review date
2026-08-17
Reviewed for
STUD
Review context
STUD runs every catalog play through third-party AI model providers, so buyer deliverables, workspace facts passed in intakes, and play outputs all cross this vendor boundary. This register covers the five GenAI IP risk categories for the draft inference agreement before signature, so counsel review starts from a complete clause map instead of a raw contract read.
Ip clauses
| Category | Clause excerpt | Contract section ref | Status | Note |
|---|---|---|---|---|
| training-data-license | Provider shall not use Customer Content, including prompts, intake parameters, workspace facts, or generated deliverables, to train, fine-tune, or otherwise improve any machine learning model, except with Customer's prior written opt-in on a per-dataset basis. | Section 6.3 (Use of Customer Content) | present | Critical for STUD: buyer workspace facts flow into every run intake and must never become vendor training data. |
| indemnification | Provider shall defend and indemnify Customer against any third-party claim alleging that model outputs delivered through the Service infringe a copyright, trademark, or trade secret, provided Customer used the Service as documented and did not disable content filters. | Section 11.2 (IP Indemnity) | under-negotiation | Draft caps the indemnity at 12 months of fees; STUD is negotiating an uncapped IP indemnity because infringing play deliverables would reach paying buyers directly. |
| confidentiality-prompt | All prompts, system instructions, intake data, and retrieval context submitted by Customer are Confidential Information of Customer; Provider shall not log them beyond 30 days, disclose them, or make them accessible to other customers or personnel without a need to know. | Section 8.1 (Confidentiality of Inputs) | present | Covers the intake payloads that carry a buyer's stated facts (pricing, pipeline, financials) into each play run. |
| output-ownership | As between the parties, Customer owns all right, title, and interest in outputs generated from Customer's inputs, and Provider assigns to Customer any interest it may hold in such outputs upon generation, to the extent permitted by law. | Section 6.1 (Output Ownership) | present | STUD's terms pass deliverable ownership through to the commissioning buyer, so upstream assignment must be clean. |
| ai-use-disclosure | Customer may disclose Provider's identity and the fact that deliverables are AI-generated to Customer's end users; Provider imposes no restriction on Customer's disclosure of AI involvement, and Customer remains responsible for its own disclosure obligations to end users. | Section 14.4 (Disclosure of AI Use) | present | STUD discloses agent authorship on every play page, so the contract must not restrict that disclosure. |
Opens soon
STUD opens runs for everyone soon, on every play at once.
Get early access to STUD the day it goes live.