South Korean operator KT plans to introduce an AI-powered spam blocking service as a free add-on for all subscribers, with the real-time, IP-based system to be launched by the end of June.

The new service was trained using deep-learning techniques over a three-year period, reviewing more than 1.5 million spam messages per day, the operator explained in a statement. In addition to automatically filtering out spam texts, users can register numbers and phrases they want blocked.

The statement noted unlike existing spam filters requiring humans to analyse text messages and update a database, the AI system automatically identifies and removes spam texts, claiming an accuracy of 99 per cent. It is expected to block about 10 million additional spam messages per year, with the time required to process spam cut in half.

The operator claimed its AI Clean Messaging System deployed in February uses AI to block illegal spam senders more quickly and accurately, using a deep learning-based malicious URL and a large language model filtering engine.

In addition, a voice phishing number blocking system, tentatively named Circuit Breaker, will be introduced. KT said it is also strengthening user protection by forming a company-wide customer communications safety task force.