Search the Atlas

Search risks, controls, and glossary terms

Non-AgenticPrivacy & Data Governance

Purpose Limitation

Explanation

Purpose limitation checks that customer data is used only for the specific, stated purpose for which it was collected and not quietly repurposed — for example, transaction data gathered to process payments must not be reused for marketing or an unrelated model without a fresh basis. It matters immediately because purpose creep is a direct violation of RBI Safety/Resilience expectations, NIST MANAGE 3.1 and ISO Clause 6.1.2, and erodes customer trust. It is measured by the Purpose Deviation Rate = Count(Data_Uses_Outside_Purpose) / Count(Total_Data_Uses) for customer transaction data — the fraction of all data uses that fell outside the declared purpose. To implement, tag each dataset with its authorised purpose(s) and instrument the access/processing layer across Agentic, Document Extraction and RAG flows to record every use with its declared purpose, comparing actual use against the registered purpose; log each data use, its purpose tag and any out-of-purpose access as evidence. The threshold is 0% — zero deviations are tolerated. Any out-of-purpose use is a breach that triggers immediate blocking of that use, an incident review and, where data was already processed, remediation, with the purpose-tagging and access controls re-examined to prevent recurrence.

Metric calculation

Deviation Rate = Count(Data_Uses_Outside_Purpose) / Count(Total_Data_Uses) For customer transaction data.

Risks mitigated

1