> For the complete documentation index, see [llms.txt](https://docs.cortex.aic.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.cortex.aic.io/workflow-guides/assurance/processing-activities.md).
# Processing Activities
**Open this area:** In the CORTEX menu, select **Assurance & governance → Processing activities**.
Retain explicit body and purpose eligibility, reviewed safeguards, and bounded independent processing decisions.
**Who uses this:** Reviewer, Administrator.\
**Open in CORTEX:** `/governance/processing`.\
**In-app guide:** `/help?topic=processing-activities`.
**Expected result:** A processing activity records the exact assessed scope and a traceable legal-owner decision, separate from source authority and data access.
## Before you start
* Choose the intended workspace and obtain the explicit processing-activity capability. Editing and reviewing also require Manager workspace authority; a role never expands the global grant.
* Collect actual controller/processor, lawful-basis, safeguards, DPIA/ethics, retention, and transfer references. Protect those details with appropriate handling.
* Identify the complete assessed resource set and exact revisions/hashes. The register does not grant access to the underlying intelligence or permission to acquire it.
## What to do
{% stepper %}
{% step %}
## Register the proposed purpose
Open Processing activities and choose Register processing activity.
The prerequisite panel shows that exact resource revisions and processing edit access are needed. Use Purpose, Safeguards, Decisions, Resources, and Review steps; each step lists missing required inputs and changes color when ready.
Enter body, jurisdiction, purpose category, and proposed regime explicitly. A police corporate activity cannot inherit Part 3; an intelligence product name cannot establish Part 4 eligibility.
An incomplete draft cannot be saved or approved.
{% endstep %}
{% step %}
## Retain accountability and assessed scope
Safeguards needs controller/processors, proposed basis, data and subject categories, sensitive-processing conditions, safeguards, retention, and transfer conditions.
Decisions needs DPIA and ethics references and rationales even where a separate assessment is not required, plus 1–5 permitted operations.
Resources needs 1–20 distinct exact revisions and SHA-256 hashes.
Review needs a reason before the final Save draft for review action.
{% endstep %}
{% step %}
## Resolve collection resources
For a collection purpose, use the source and requirement selector to apply one source and 1–19 current requirements covered by its handling.
This resolves exact revisions and hashes without manual copying; it grants no operations or legal approval.
From an approved activity, an authorised case editor can prepare a source authority case carrying the same body, jurisdiction, purpose, and resources.
{% endstep %}
{% step %}
## Obtain independent applicability review
A separately authorised reviewer who has not edited any draft opens Review applicability.
Retain the external legal-owner decision, actual UTC decision time, eligibility evidence, and clause/edition/commencement mapping.
Select a current approved UK profile and explain adoption. Set owner-approved effective, review, and expiry dates in UTC.
{% endstep %}
{% step %}
## Reassess material changes
Use Reassess activity for purpose, dataset, model, provider, or other changes.
Every amendment creates a draft and clears current approval. Prior decisions remain in history; a new independent approval is required before further controlled use.
{% endstep %}
{% step %}
## Use the purpose-specific AI decision
Report generation, retained News triage, and evaluation execution use the exact configured model, provider, prompt, and dataset purpose.
Generation and Evaluation have separate processing selections. Acquisition or source authority alone is insufficient; changed runtime or data scope requires reassessment before another model call.
{% endstep %}
{% step %}
## Suspend or close deliberately
An authorised reviewer records a reason to suspend use or close the activity.
Suspension requires reassessment and approval before reuse. Closure is final for this activity; create a successor.
Neither action deletes records or authorises disclosure.
{% endstep %}
{% step %}
## Inspect retained evidence
Choose a revision in Retained decision history to inspect its exact content, attributed decision, and hash.
Historical approval is evidence only; current use also checks access, profile, exact scope, and dates.
{% endstep %}
{% endstepper %}
## Check the result
* Is the purpose explicit rather than inferred from the organisation name?
* Does the legal owner support the selected body, purpose, jurisdiction, and regime?
* Are resource revisions and the complete processing scope current?
* Are sensitive conditions, assessments, and safeguards evidenced?
* Is the reviewer independent and are all UTC dates taken from the actual decision?
## If you get stuck
Why can I not approve this activity?
Resolve unassessed or incompatible body/purpose/regime values, supply the required legal decision and clause mapping, and select a current approved profile.
The reviewer must not have edited any retained draft. Review must be in the future, after the effective start and no later than expiry.
Why did a previously approved use stop?
Check for a later draft, suspension, closure, changed resources, due review, expired period, or unavailable profile.
Reassess the current scope; do not substitute a retained historical approval.
How do I approve restricted analytical reuse?
In the analysis editor, use Processing purpose for this reuse to inspect the current requirement and exact dataset extraction references.
An authorised activity owner records that complete scope, the purpose, and Analysis operation for independent review. Select its approved activity before saving restricted work.
Submission and an endorsing peer review recheck current approval. A changed dataset or requirement requires reassessment; retained history stays readable under normal access.
What does a restricted product need?
Inspect processing resources in the product editor after selecting citations, linked analysis, and approved report sources.
The scope includes all linked analytical evidence and the exact approved report projections. Select a current activity permitting Analysis for that purpose before saving.
Shield rechecks the original scope and keeps the decision on its derivative; an older original without a decision needs a reassessed original revision first.
Historical product reads and exports retain the recorded decision without treating it as current release approval.
How is a release assessed?
Select the current product and verified recipients in the release editor, then inspect current assessed resources.
The decision must permit Dissemination for the exact product and audience fingerprints. Submission, endorsement, approval, and publication recheck it.
Amend a changed audience or stale decision and obtain fresh review; cancellation and changes requested remain available.
Previously published packages keep their exact bytes and recorded decision, with current access and recipient checks on retrieval.
Does an approval itself authorise collection or release?
No. It records processing applicability for an exact bounded scope.
Each connected workflow still needs its own current permissions, handling, source authority, and approval checks.
Does this automatically handle rights requests or erase records?
No. This activity workflow neither discloses nor destroys holdings.
Rights-case and downstream correction workflows are separate responsibilities.
## Responsibility and limits
* Read, edit, review, and use are separate explicit capabilities. Privacy administration does not grant unrestricted intelligence access or source authority.
* No universal statutory deadline is supplied. New use stops at the review/expiry boundary and also depends on the adopted profile remaining current.
* An uncertain registration can be retried with the same values on the same page. A changed registration conflicts; reconcile the retained record before creating another.
* Failed validation or concurrency retains the entered draft. Access loss clears sensitive page content. Do not put legal-case details into support logs.
* Legal-owner commencement and organisational acceptance remain open until actual evidence is supplied. Synthetic tests and application forms do not complete those decisions.
**Reference and scope:** F14 processing applicability. Actual legal-owner evidence, jurisdiction, source editions, and commencement remain required; technical permissions do not establish legal powers.
## Related workflows
* [Record and review external collection authority](file:///7397340/collection/authority-cases.md)
* [Record authority before new source use](file:///7397340/collection/source-authority.md)
* [Preserve an exact revision and manage holds](broken://pages/0185ca5a2331760fb4c59665b8dde7f4d80e0fc6)
* [Manage accounts, grants and workspace membership](broken://pages/99c8701bc497eefb32769b668198b1f9f5fc6a93)
* [Maintain a system control and evidence baseline](broken://pages/699956df97c18fc4ddd8894884fea67a7b6e90ce)
* [Record UK processing paths and approved uses](broken://pages/d1a5ce8a791f7e21103f6f699e97b63ddcffb253)
* [Develop and independently review an assessment](file:///7397340/analyse/analysis.md)
* [Author an intelligence Product and inspect its exports](file:///7397340/product/products.md)
* [Review, approve and publish an exact internal release](file:///7397340/product/release.md)
* [Handle rights requests and retain response evidence](broken://pages/a35e2f0757e5c31ace450534e7af7b26f6940365)
* [Assess evidence quality and retain defect actions](file:///7397340/analyse/evidence-quality.md)
* [Govern AI use cases and exact data purposes](broken://pages/c1669122986904b8ceb9c274b7824d3658d7d94b)
* [Evaluate AI and restore suspended use](broken://pages/391a874ae052db4fcb4ee584bd7dc26eead47228)