False positive
When a detection system raises an alarm about something innocent.
A false positive is a wrong yes. A spam filter that bins a real invoice, a security scanner flagging safe code, an AI detector accusing a human writer: all false positives. The opposite, missing something real, is a false negative, and every detection system trades one against the other. Tune it to catch more, and you accuse more innocents. Tune it to accuse fewer, and more slips through.
The rate is usually reported as a percentage, which makes small numbers look harmless. It helps to convert them. A false positive rate of 0.01 percent sounds like nothing until you run a million documents through it and produce a hundred wrongly accused people, each of whom has to prove a negative.
-
Anthropic loosens Claude Fable 5's biology filter, cutting blocked questions by 85 percent
-
OpenAI open-sources its vulnerability-hunting tool, and hands it to anyone with a terminal
-
Pangram says its new detector is wrong once every 24,000 documents. That number deserves a closer look
-
AI Cheating Is Getting Easier to Do and Harder to Prove