False Positive

False positives are a cost of any detection system: a DLP policy that catches credit-card patterns will fire on test strings and documentation; a PII classifier will tag non-personal numeric sequences; a SIEM rule will alert on benign anomalies. The signal-to-noise ratio determines whether the detection layer is actionable or ignored, and well-tuned programmes track false-positive rate as a core KPI alongside true-positive rate.

For data classification specifically, false positives inflate the apparent sensitive-data footprint and create wasted DSR fulfilment work (searching for a subject's data in stores that only matched a pattern, not an actual record). Modern classifiers combine pattern matching with context (field names, surrounding content), ML-trained category detection, and dictionary matches against validated identifiers to lower the false-positive rate below what pure regex can achieve. The complement — false negatives — are usually more dangerous (missing real personal data) but less visible until an audit or breach reveals the gap.