Behavior-based detection is an approach to security and fraud control that identifies threats by what they do rather than by what they look like. Instead of comparing a file against a list of known malware fingerprints, as signature-based detection does, a behavior-based tool watches actions such as a program encrypting many files at once, contacting unusual servers or trying to read stored passwords, and blocks or flags it when the pattern looks harmful.
The same idea underlies much of modern antivirus and endpoint protection, and it is used in banking and crypto in the form of anomaly detection and transaction monitoring: a login from a new device, an unusual payment size or a sudden change in typing pattern can trigger extra checks. Its strength is catching new threats that have no signature yet; its weakness is a higher rate of false alarms, which systems try to reduce with tuned rules, heuristics and machine learning.
Ref 1050000669