Naver Upgrades AI Cleanbot to Block Malicious Comments

SEOUL, South Korea — Naver Corp. has announced the deployment of AI Cleanbot 3.0, an upgraded artificial intelligence system designed to detect and block malicious comments.

The updated system targets comments that encourage disregard for life, including content related to suicide, death, and self-harm.

The company said the software is also programmed to filter out mockery, derogatory language, and hate speech targeting victims of accidents and their families.

AI Cleanbot 3.0 evaluates user comments by analyzing them alongside the corresponding news article's headline and body text.

Naver said integrating article context into the algorithmic assessment improves the system's ability to accurately detect malicious intent.

"We are continuously working to improve the performance of Cleanbot to detect newly emerged hate, derogatory, and discriminatory expressions, as well as profanity and slang," said Su-hyang Kim, a leader at Naver.

The system's development began with its first iteration in November 2019, which used keyword-based filtering.

Naver adopted deep learning for context analysis in 2020. Subsequent updates addressed profanity modifications, symbols used to bypass filters, evolving internet slang, and incorporated the Korea Internet Self-Governance Organization's guidelines on hate speech in 2023.

Separately, Naver this year implemented platform policies that disable comments on sections covering politics and elections.

The platform also automatically deactivates the comment section on individual articles when the volume of malicious comments exceeds a specified threshold.