open-weight-models

How local AI tools protect privacy: on-device inference keeps prompts off servers, quantization makes it fit, and telemetry, storage, and exposure matter.