QuestionLegal ServicesAI GovernanceData Protection

Do AI contracts stop client data being used for model training?

18 August 2026
Answered by Rohit Parmar-Mistry

Short answer

Before putting client material into an AI tool, check the terms for that product tier. Confirm whether prompts, files and outputs can be used for training, how long they are retained, who can access them and what deletion rights apply.

Detailed answer

Why the contract matters before client data reaches an AI tool

Before putting client material into an AI tool, check the terms for that product tier. Confirm whether prompts, files and outputs can be used for training, how long they are retained, who can access them and what deletion rights apply.

Keep a control trail that records what data leaves the firm, where it is processed, how long it is retained, who can review it, which subprocessors touch it, and what happens when you ask for deletion.

Set the contract boundary for the proposed workflow

If approval depends on excluding client material from training, verify the binding commitment for the exact feature, tier and settings. Record its scope and exceptions before approval. Assess any redacted alternative separately, and review retention, access and onward use alongside the training term.

For legal work, assess confidentiality, privilege, data protection, applicable professional duties and matter-quality review before approving sensitive material for an AI system. Check the contract and record the decision for the actual data flow; the supplier's role and terms need to be assessed for that use.

Check your AI tool risk and efficiency controls

What to Check in the Proposed Contract Commitment

A useful contract should be specific. Look for language that says customer content, prompts, files, embeddings and outputs are not used to train foundation models or improve shared services. The wording should also cover temporary retention, abuse monitoring, support access, human review, backups and logs. A narrow promise about model training is helpful, but it does not answer every confidentiality risk.

The firm should also check whether the no-training commitment applies by default or only after an admin setting is changed. If the protection depends on an enterprise tier, a private workspace, a zero data retention option or a particular regional setting, that needs to be documented in the internal approval record.

How to turn the contract into a usable control

Start with a short intake checklist for each AI-enabled workflow. Record the supplier, intended use, data categories, retention period, subprocessors, transfer basis, audit evidence and reviewer owner. Then decide which uses are allowed, restricted or prohibited. For example, public research summaries may be allowed, anonymised precedent analysis may need review, and raw client bundles may be prohibited unless a stronger private deployment is in place.

Staff also need plain rules, not a long policy that nobody uses. Tell people which approved tools they may use, what data must be removed first, when a human must review the output, and where evidence of the check is stored.

Keep AI governance evidence current

What to do if the terms are unclear

List unresolved training, human-access, retention, location, recipient and deletion questions and request product-specific evidence. Do not infer protection from silence. Any alternative input set needs its own review; a non-confidential label alone does not approve the use.

For higher-risk work, consider a private deployment, a legal-specific platform with stronger contractual terms, or an internal workflow that strips personal and client-identifying data before any AI processing. The right answer depends on matter sensitivity, user training, integration design and the quality review process around the output.

Conclusion

List the review conditions for retention, access, recipients, transfers, output checks and evidence for the proposed use. A training commitment answers only the use it actually covers. Keep material outside approval where a necessary condition remains unresolved.

Build a controlled AI implementation plan

Need a safer AI route for your firm?

If this article reflects a live decision in your practice, the useful next step is to map the workflows, confidentiality risks, supplier controls and governance gaps before tools spread informally.

Frequently asked questions

Is a vendor FAQ enough evidence?

No. Use the contract, data processing terms, security documents and admin settings as the evidence base. A public FAQ can support the review, but it should not be the only control.

Can staff use public AI tools with anonymised information?

Sometimes, but anonymisation must be real and repeatable. Remove client names, matter facts, unique identifiers and commercially sensitive details, then require human review of the output.

What if an enterprise plan says it does not train on our data?

Check whether that protection is contractual, whether it applies to all features, and whether any settings must be enabled. Record the evidence and revisit it when the supplier changes terms.

Who should own the decision?

Ownership should sit with a named business and risk owner, supported by IT, data protection and matter-quality reviewers. Record who can approve the use and who must revisit it when the workflow or supplier terms change.

Discuss your next step

Talk through your situation with Rohit, or take the assessment if you need a first read.