Home / Privacy and regulated data
Redaction for Regulated Data
Here the obligation is standing rather than triggered by a request. The data is regulated wherever it sits, and the question an auditor asks is what proportion of the framework's identifier set you actually caught.
Use Cases
HIPAA and PHI
Compliance
Redact protected health information from patient recordings, clinical session videos, billing calls and medical documents.
Learn moreMedical and Clinical Records
Documents
Detect and redact PHI from patient records, clinical trial documents, scanned files and telehealth recordings before sharing with sponsors, monitors, regulators and CROs.
Learn morePCI DSS Call Recordings
Audio
Remove card numbers, CVVs and cardholder data from call recordings automatically, without losing the recording's value for QA and training.
Learn moreContact Centre Recordings
Audio
Detect and remove sensitive data spoken in recorded calls. Speaker separation keeps agent responses intact while customer information is redacted, thousands of recordings per batch.
Learn moreScreen and Session Recordings
Video
Detect and redact PII visible in recorded meetings, QA sessions and desktop captures, including on-screen text in dashboards and application interfaces.
Learn moreChallenges
What These Share
Coverage is the number that matters
When a job runs against a framework, the dashboard reports how much of that framework's identifier set was actually redacted, across twelve frameworks each with its own class set.
Clinical text is not ordinary text
Identifiers in a medical record take forms they do not take in a contract, which is why clinical content runs through a model trained on the i2b2 clinical corpus rather than a general PII recogniser.
Who said it decides what to cut
In a recorded call the customer's card number has to go and the agent's script has to stay. Speaker separation is what makes that distinction available to the redaction.
Regulated volume is continuous
These are not one-off releases but a standing flow, which makes automated bulk processing and API-driven pipelines the normal mode rather than the exception.
Running It Continuously
Connect the source
A watched location, a bucket or an API call, so files are taken as they are produced rather than gathered by hand.
Apply a policy
A saved template fixes which classes are detected, which framework coverage is measured against, and which masking style is used, so every file in the flow is treated identically.
Process without an operator
Headless bulk processing runs the policy across the queue, permissioned per format.
Report coverage
The dashboard states what proportion of the framework's identifier set was redacted, per job and across the portal.
Export the record
The filtered redaction record exports as CSV for the audit file.
Where This Fits
Regulated content spans formats. Document redaction handles clinical records and scanned forms, audio redaction the recorded calls with speakers told apart, and DICOM redaction the imaging studies where identifiers sit in both the pixels and the header tags. Detection lists what is recognised before anything is masked, and security covers how the unredacted original is protected while it is held.
Run It Against Your Own Framework
Send us a sample and we will report coverage against the framework you are audited on.