---
title: 'False positive'
description: 'A negative case that a classifier or detector incorrectly labels as positive.'
canonical_url: 'https://darkfactory.dev/glossary/false-positive'
markdown_url: 'https://darkfactory.dev/glossary/false-positive.md'
collection: glossary
date_published: '2026-08-03T00:00:00-04:00'
date_modified: '2026-08-26T00:00:00-04:00'
---

# False positive


## Definition

A false positive is a negative case that a classifier or detector incorrectly labels as positive. A legitimate transaction blocked as fraud, safe code reported as vulnerable, and a human-written essay flagged as AI-generated are all false positives. Which outcome counts as "positive" is a convention and must be stated.

Do not confuse the count with the **false positive rate**. The rate is `false positives / (false positives + true negatives)`, so its denominator is all actually negative cases. Precision uses a different denominator: all predicted positive cases.

The right threshold depends on consequences. A low threshold may catch more real problems while sending more harmless cases to review. Teams should report both sides of that tradeoff and identify who bears the cost of a false alarm.

## Distinguish it from nearby terms

- A false negative is a positive case incorrectly labeled negative.
- A true positive is a positive case correctly detected.
- A false discovery is a positive prediction that proves wrong. Its rate is related to precision, not to the false positive rate's denominator.

## Check your understanding

A security scanner flags 20 of 1,000 safe files and 80 of 100 unsafe files. What is its false positive rate, and what other rate describes the unsafe files it missed?

## Related terms

- [False negative](https://darkfactory.dev/glossary/false-negative)
- [Precision](https://darkfactory.dev/glossary/precision)

## Evidence and further reading

- [NIST AI Resource Center Glossary](https://airc.nist.gov/glossary/)
- [Google Machine Learning Glossary](https://developers.google.com/machine-learning/glossary/)