Personal data in Jira and Confluence: answering a data subject request
Also published on Atlassian Community
In this guide
A former customer writes in: "Please erase all personal data you hold about me." The CRM and the mailing list have clear owners. Then someone remembers the support tickets, the onboarding page in Confluence with a scan of a passport, and the spreadsheet attached to an incident. Nobody owns those.
LeakRadar is a secret and personal data scanner for Jira and Confluence Cloud from Radrly. This guide follows one data subject request from the day it arrives to the day you answer, with the native Atlassian steps first and the app near the end. It is not legal advice; what you must erase and what you may keep is for your DPO and your counsel.
Day 0: log the request and start the clock
Under the GDPR you have one month to respond, extendable in some cases. Log the request in a restricted place, and write down the identifiers you will search for: email addresses, phone numbers, customer IDs, and any national ID the person gave you. A name alone finds too much and misses too much, so treat it as a last pass.
Days 1 to 5: find every place
Jira. A JQL search such as text ~ "jane.doe@example.com" covers the current text of summaries, descriptions, comments and text fields. It does not read field history or attachments. Run it once per identifier, and repeat with Confluence search for pages, blog posts and comments.
Field history. If someone already edited the value out, the History tab of the work item still shows the old text to anyone who can browse it. Search does not find it there.
Page versions. Confluence keeps older versions of every page. A value removed today is still in every version saved since it was added.
Attachments. Screenshots, PDFs and spreadsheets are where identity documents end up. Native search does not open them.
Accounts. If the person also has an Atlassian account, deleting a managed account removes profile data such as name and email from Atlassian account services, and after 14 days the person shows as Former user. Atlassian is explicit that personal data typed into pages and work items, by them or by anyone else, is not deleted.
With Atlassian Guard Premium, content scanning detects identity data such as US Social Security Numbers when content is created or updated, and an organization admin can redact a finding, including Jira work item history.
Days 6 to 20: erase what you decided to erase
Edit the current text in each work item, page and comment. For Confluence, delete the older versions that still contain the data from Version history; per Atlassian's docs, deleting a page version is permanent. For Jira history without Guard Premium, Atlassian's own advice is to edit the field, clone the work item and delete the original, which gives the work item a new key, so check what the clone carries over before you delete. Delete or replace attachments.
Days 21 to 30: record and answer
Keep a record of what was found, what was erased and what was kept and why, without copying the personal data into the record itself. Then answer the person.
Where LeakRadar fits
LeakRadar scans Jira summaries, descriptions, comments, the latest 20 history entries of summary, description and environment, and attachments, plus Confluence pages, blog posts, comments, attachments and up to 15 older versions of a page. The detectors that matter here include phone numbers, email addresses (off until you enable them), IBAN and 19 national IDs such as PESEL, Steuer-ID, BSN, NINO and SSN, with checksums where they exist. The GDPR framework pack turns the relevant detectors on in one step. Personal data is fully masked in every finding.
The request goes to Audit > Erasure. You enter an email address, phone number or national ID. LeakRadar hashes it with the installation key and lists the findings that already carry that hash. The typed value is not stored or written to the audit log, and the lookup is kept for 15 minutes as hashes only.
Erasure then redacts the current text as the administrator who confirms it, can purge older Confluence versions (off by default) and delete attachments, and lists the Jira history locations that apps cannot delete, so you know exactly what to hand to Guard or to the clone and delete route. The erasure record stores counts (redacted, purged, history and failed) and a SHA-256 hash, which becomes the next link in the evidence chain. That record is the Day 30 evidence, and it holds no personal data. The Rovo agent never accepts the value and points to this screen.
The limits are worth stating plainly. The lookup only matches values LeakRadar has already found, so a name, a customer ID or text outside the scanned locations will not match. Resolved and false positive findings are left out. Apps cannot remove Jira field history. DSAR lookup and erasure are part of Standard, and signed evidence packs are in Advanced.
Quick answers
Does native Jira search find personal data in field history and attachments?
No. A JQL search such as text ~ "jane.doe@example.com" covers the current text of summaries, descriptions, comments and text fields. It does not read field history or attachments, and Confluence keeps older page versions that still contain a value removed today.
Does deleting a managed Atlassian account remove the person's data from pages and work items?
No. Deleting a managed account removes profile data such as name and email from Atlassian account services, and after 14 days the person shows as Former user. Atlassian is explicit that personal data typed into pages and work items is not deleted.
How do I remove personal data from Jira field history without Guard Premium?
Atlassian's advice is to edit the field, clone the work item and delete the original, which gives the work item a new key, so check what the clone carries over before you delete. With Guard Premium an organization admin can redact a finding, including Jira work item history.
How does LeakRadar handle a data subject request?
On Audit > Erasure you enter an email address, phone number or national ID. LeakRadar hashes it with the installation key and lists the findings that already carry that hash. Erasure redacts the current text, can purge older Confluence versions (off by default) and delete attachments, lists the Jira history locations apps cannot delete, and stores a record with counts and a SHA-256 hash. The lookup only matches values LeakRadar has already found. DSAR lookup and erasure are part of Standard.