Ahead of the 9th nationwide local elections on June 3, major platforms such as Naver, Daum and TikTok are strengthening systems to provide election information and stepping up responses to false information and comment abuse. While companies are using upgraded artificial intelligence-based detection systems and links to authoritative election information as key tools, they differ in how they build special pages and respond.

Naver does not provide comments below political section articles; steps up detection with Cleanbot 3.0

Naver opened a special local election page on May 8. It provides media picks, election news by region, opinion polls, and key election schedules, general information and past turnout rates. On election day it will run a "special election live" section that provides joint exit poll results from three broadcasters along with forecast poll results by media outlets.

To prevent comment abuse, it has not provided comments below the main text of articles in the politics and election section from March 19 to June 3. This is seen as a measure to reduce concerns about artificially shaping public opinion through organised comment activity during the election period. From April 23 it has also been implementing a policy to automatically disable comment services on articles when malicious comments exceed certain standards.

Naver upgraded its AI-based detection system, Cleanbot, to version 3.0. It improves accuracy in detecting malicious comments by analysing not only comment content but also article headlines and bodies. It works with the Korea Internet Self-Governance Organization (KISO) and the National Election Commission to handle requests to delete comments related to candidates, and it also runs a 24-hour monitoring system and hotline.

Daum visualises election context with an AI timeline; detects abnormal abuse patterns

Daum opened a special election page in late March. Sections are currently organised into AI keywords, interviews, opinion polls, in-depth, spotlight regions, by-elections, metropolitan, local and lawmakers and superintendents of education, timeline, and regional race outlooks.

Its "AI keywords" publishes weekly the top 20 local governments by mention index and provides AI analysis of policy keywords by region. The "timeline" visualises key remarks, core issues and past histories of candidates in chronological order so users can grasp the context of the election landscape on one screen. It plans to feature the special page on Daum's front page and on the mobile home and news tab from May 29, when early voting begins.

For false information and abuse detection, it applies an AI malicious-comment filtering system called Safebot. It also runs "dynamic weighting control" that detects five conditions — sudden traffic concentration, simultaneous reaction patterns, abuse account detection, mass-reply abuse and abnormal patterns by time period — and lowers exposure weighting in real time when organised intervention is suspected. On real-time trends, it applies a principle of excluding keywords for candidates and related figures during the period in which election influence activities are banned, reducing the possibility of opinion manipulation in advance.

TikTok builds an official election commission account and in-app election centre; requires AIGC labels

TikTok Korea held an online briefing on May 20 and disclosed its fairness policy for the local elections. The National Election Commission has opened an official TikTok account ahead of the elections to provide election schedules, voting procedures and guidance on early voting. TikTok said some of the commission's posted content has already recorded more than 20 million to 50 million views. When users search for election-related keywords such as "local elections", "early voting" and "polling place", a banner 안내 appears for an "in-app election centre" that links to the commission's official information. A TikTok official explained that it is structured to connect users to information from authoritative institutions rather than providing the platform's own interpretation of election information.

To respond to false information, it uses both automated detection technology and human review and works with global fact-check organisation Lead Stories to review factual accuracy of Korea-related content. Based on its latest quarterly transparency report, 99 percent of deleted content was handled proactively, and 86 percent of that was removed before it generated views. It operates a dedicated cooperation channel with the election commission's cyber election crime response team, and decides whether to take action after reviewing reported content for violations of community guidelines and domestic law.

For AI-generated content (AIGC), it requires creators to indicate that content was generated by AI when it appears factual. When TikTok identifies AI generation through cooperation with content provenance authentication systems such as C2PA, it sometimes attaches labels itself. Regardless of whether content is labeled, if it violates guidelines or domestic law it is deleted or restricted in exposure. Accounts related to governments, politicians and political parties are not allowed to run political ads, raise election funds or participate in creator monetisation programmes.

Responses differ by platform depending on service structure. Naver and Daum are putting emphasis on access to election information and comment management in a portal environment centred on news and comments. TikTok, by contrast, is focusing on linking to the election commission's official information and responding to AIGC, considering that false information can spread quickly in a short-form video and search-and-recommendation environment. This local election is expected to serve as an opportunity to assess major platforms' election information systems and capabilities to respond to false information.