30 Days Gen AI Risk Trial -Start Now
Skip to main content

Developer data protection

  • Claude

Choose the right Claude Code account

Approve the account and sign-in arrangement used for company code, not only the Claude Code application. Anthropic’s current documentation distinguishes consumer training preferences, commercial terms, retention options and separate feedback-sharing flows. Confirm which arrangement governs the intended session, then record the organisation’s permitted code and purpose.

For IT procurement and engineering leads

Aona field notesD11
Account decision
Sign-in determines the review

Record the applicable terms before approving company code.

Dated provider categories plus a fictional decision scenario. No real account, agreement or approval is verified.

01

Identify the actual sign-in arrangement

A developer can use the same application under a personal subscription, an organisation-managed arrangement or a provider/API route. The appearance of the editor is not evidence that the intended commercial terms apply. Record the account owner, plan, provider and authentication method using non-secret administrative evidence.

Do not collect keys or authentication files for this review. A reference to the approved account and its owner is enough. If a developer changes the sign-in method or introduces a different provider, treat that as a new policy question rather than assuming the earlier application approval follows automatically.

Source context: Claude Code: Data usage

02

Compare the training commitments

For consumer Free, Pro and Max accounts, Anthropic documents a user choice about allowing data to improve future models, including Claude Code use. Under commercial terms, it states that code and prompts are not used to train generative models unless the customer chooses to provide data for model improvement.

The distinction is about the applicable terms and choices, not whether the employer permits disclosure. Company code may still be restricted even where a no-training commitment applies. Record any organisational opt-in programme separately, along with the owner authorised to make that choice.

Source context: Claude Code: Data usage

03

Record retention for the selected route

The data-usage documentation reviewed on 21 September 2026 describes consumer retention of five years when model-improvement use is allowed and 30 days when it is not. It describes a standard 30-day commercial retention period and qualified zero-data-retention arrangements, which require specific enablement rather than being included automatically in every Enterprise plan.

Provider arrangements, agreements and features can change the relevant handling. Record the source and arrangement that apply to the actual session. Keep a blank field when the organisation has not confirmed a term; do not convert an unknown value into a claim of zero storage.

Record retention for the selected route
ArrangementDocumented questionOrganisation evidence
Consumer accountModel-improvement preference and retentionActual account setting and permitted use
Commercial arrangementTerms and any explicit data-use opt-inContract/account owner confirmation
Qualified retention variationSpecific enablement and scopeWritten agreement and covered features
Third-party provider routeWhich provider’s handling applies?Selected route and applicable terms

Source context: Claude Code: Data usage

04

Keep support sharing and local records visible

Normal session policy is only part of the decision. Feedback reports and optional transcript sharing have separate documented flows and retention. A developer can create a further disclosure by sending a support bundle even when ordinary coding use is approved. Link the support-upload procedure to the account approval instead of duplicating it here.

Also identify where local session records and exported material may remain. A provider policy is not a complete inventory of files on the employee’s device or records held by another connected service. Assign an owner to any additional record required by the organisation’s handling policy.

Source context: Claude Code: Data usage

05

Approve code use within a clear scope

Use the worksheet to record the account, applicable commitments, permitted code classes and tasks, and who can change the arrangement. If the evidence is incomplete, keep restricted material out of the workflow while the responsible owner resolves it. A personal account is not made organisation-approved merely by expensing its subscription.

Revisit the approval when sign-in, plan, provider, data-use preference or feedback practice changes. File and shell permissions remain a separate review. Choosing an appropriate account does not prevent a local tool from reading material the task should not include.

Put it into practice

Claude Code account and policy approval

Pair the actual sign-in arrangement with applicable training, retention and code-use decisions.

Dated provider categories plus a fictional decision scenario. No real account, agreement or approval is verified.

Claude Code account and policy approval
CategoryDated documentationReview implication
Consumer, improvement allowedFive-year retention describedEmployer permission still required
Consumer, improvement disabled30-day retention describedConfirm actual preference and task
Commercial standardNo training absent opt-in; standard 30 daysConfirm agreement and provider route
Qualified ZDRSpecific enablement requiredDo not infer it from plan name
Feedback report (/feedback)Five-year path, including /bug and /shareReview the exceptional upload

Work through your review

Use the checks to organise the evidence you need. Your selections stay in this tab.

0 of 3 reviewed

Example files for this task

Keep the source material and the instructions together. You can also download the complete worksheet or matrix as CSV.

README.mdInspect
# Claude Code account review

This pack contains no credentials or active configuration. It does not sign in, contact a provider or approve a disclosure.

Record the actual account/provider and current policy sources in account-review.csv. Use the decision record for employer-owned code. For support disclosures use the separate D09 guide, and for local file permissions use D05. Do not infer zero retention or commercial coverage from the product name alone.

## Guide and sources

Canonical guide: https://aona.ai/resources/guides/claude-code-personal-commercial-data-policy/
Source review: 2026-09-21
- Claude Code: Data usage: https://code.claude.com/docs/en/data-usage

The pack also contains a populated dated policy-category table and a clearly fictional account decision. Those examples are separate from the blank actual-account review.
Download README.md
account-review.csvInspect
field,current_evidence,source,owner,status
Account owner,TO_RECORD,,ASSIGN,UNVERIFIED
Plan and provider,TO_RECORD,,ASSIGN,UNVERIFIED
Authentication method without secret values,TO_RECORD,,ASSIGN,UNVERIFIED
Training preference or applicable commitment,TO_RECORD,,ASSIGN,UNVERIFIED
Explicit improvement programme opt-in,TO_RECORD,,ASSIGN,UNVERIFIED
Standard retention,TO_RECORD,,ASSIGN,UNVERIFIED
Qualified retention variation and enablement,TO_RECORD,,ASSIGN,UNVERIFIED
Feedback and transcript-sharing process,TO_RECORD,,ASSIGN,UNVERIFIED
Local and exported records,TO_RECORD,,ASSIGN,UNVERIFIED
Download account-review.csv
code-use-decision.mdInspect
# Company code use decision

Account/provider reviewed: ____________________
Applicable terms and review date: ____________________
Permitted code classes and purposes: ____________________
Restricted material: ____________________
Owner authorised to change data-use choices: ____________________
Feedback-sharing procedure: ____________________
Retention evidence and unresolved questions: ____________________
Decision: NOT YET REVIEWED
Recheck after sign-in, plan, provider, preference or agreement change.
Download code-use-decision.md
documented-account-categories.csvInspect
checked_at,category,documented_policy,important_boundary,source
2026-09-21,Consumer improvement enabled,Data may improve models and five-year retention described,Actual preference and employer permission to confirm,https://code.claude.com/docs/en/data-usage
2026-09-21,Consumer improvement disabled,30-day retention described,Confirm actual selected account and preference,https://code.claude.com/docs/en/data-usage
2026-09-21,Commercial standard,No training unless opted in and standard 30-day retention described,Provider agreement and feature scope to confirm,https://code.claude.com/docs/en/data-usage
2026-09-21,Qualified zero retention,Specific organisation enablement required,Not automatic with standard Enterprise,https://code.claude.com/docs/en/data-usage
2026-09-21,Feedback report (/feedback plus /bug and /share),Separate five-year retention path,Not ordinary session retention,https://code.claude.com/docs/en/data-usage
Download documented-account-categories.csv
fictional-account-decision.mdInspect
# Fictional account decision

SYNTHETIC SCENARIO ONLY. No real account or approval exists.

Hypothetical task: explain a generic toy sum function. No code file is supplied in this account-policy pack.
Candidate A: a fictional employee-funded consumer account with improvement preference unverified.
Candidate B: a fictional organisation-managed commercial account.
Evidence considered: the dated categories in documented-account-categories.csv; actual contract, provider route and account state remain unverified.
Illustrative outcome: select Candidate B for the synthetic exercise only after the account owner verifies the arrangement. Withhold all real employer code until its data owner approves the specific service and purpose.
Feedback condition: no source/history submission without the separate support review.
Unresolved: actual retention variation and any data-use opt-in.
This teaches the decision structure and does not grant permission to any organisation.
Download fictional-account-decision.md

Before you proceed

Keep these distinctions clear

Approving the app rather than the account
Two sessions in the same client can use different sign-in arrangements or terms.
Treating Enterprise as automatic zero retention
Verify the specific agreement, enablement and feature scope documented for the organisation.

Apply it to employee AI use

Bring your actual data path.

Aona can help evaluate supported employee endpoint input paths alongside the organisation’s account policy.

Aona does not change Anthropic’s terms or supply unverified account restrictions. Local file access and provider handling require their own evidence.

Bring the approved account arrangement and one synthetic coding task to a scoped endpoint review.

Review your use case

FAQ

Questions for this decision

Does a paid personal plan automatically use commercial terms for company code?
Do not assume that. Identify the actual plan, provider and sign-in arrangement and review the applicable terms. Employer approval to disclose code is a separate decision.
Is zero retention included in every Claude Enterprise account?
The current Claude Code documentation says qualified zero-data-retention arrangements require specific enablement and are not simply included in the standard Enterprise plan. Confirm the organisation’s agreement and covered features.
Can commercial data still be shared through a feedback report?
Feedback and transcript-sharing choices are separately documented flows. Follow the approved support-upload review before sending code or history, even when ordinary coding use is permitted.
Does the account review replace .env permission tests?
No. Account terms concern processing and use of data. Local permissions concern what material the client or its tools can access. Review both boundaries independently.

Evidence behind the guide

Sources and scope

Prepared by Aona. Sources checked 2026-09-21. The cited material supports the specific points below; it does not certify a product or your use case.

  1. Claude Code: Data usage

    Documents consumer/commercial training policy, retention options and separate feedback flows.

    vendor · checked 2026-09-21
Claude Code personal versus commercial data policies