AI · SEMICONDUCTOR · FACT CHECK
SKD WIREMostly True
A majority of user reports concerning harmful content on AI chatbot services remain stuck in the 'under review' stage, and some cases have gone unaddressed for more than a year, according to findings released by civil society researchers, fueling criticism that operators are failing to respond to dangerous outputs in a timely manner.
The report, archived by fact-checking researchers and drawing on nine primary sources, found that 64 percent of harmful-content reports filed against AI chatbot services were still classified as 'under review' rather than resolved. Among the unresolved cases, the longest had remained unaddressed for 494 days — more than 16 months after the initial report.
Civil society organizations, including the Korea Communications Standards Commission-affiliated channels referenced in the report's sources, have raised concerns that prolonged inaction leaves users exposed to harmful chatbot outputs such as offensive or dangerous content. The figures suggest that reporting mechanisms exist on paper but may lack effective follow-up procedures to resolve flagged cases.
The findings have prompted calls for AI service operators to establish clearer timelines and disclosure practices for handling user reports of harmful content, and for regulators to scrutinize whether current review processes meet adequate standards of care.
Details such as which specific chatbot services were covered, how many total reports were examined, and whether regulators plan formal enforcement have not been confirmed in the available material, leaving the full scope of the problem open.
The finding that 64 percent of harmful-content reports on AI chatbots remain 'under review,' including one left pending for 494 days, is substantially supported by the report's sources, though some operational details remain unresolved — making the claim, on balance, Mostly True.
