Local machine learning is a model that runs on the device itself, or in a nearby trusted environment, instead of sending data to a remote server for every decision. In cyber security, this matters because it can inspect content quickly, preserve privacy, and keep some analysis inside the trust boundary of the app or endpoint.
Defenders use local ML for tasks such as scam detection, spam filtering, malware triage, and anomaly spotting. Because decisions happen close to the user, warnings can appear at the exact moment a risky action is about to happen. The tradeoff is that local models are constrained by device resources and may be easier to evade with subtle wording changes, so they can produce false positives or miss new attack patterns. For that reason, local ML is usually a layer of defense, not a complete security control.