AI-generated misinformation causes social panic in China
Reportedly, AI-generated misinformation spread widely in China in 2025, causing public panic through false reports of disasters, deaths, and disease outbreaks. Authorities attributed the spread to both deliberate misuse by individuals seeking engagement and AI hallucination—where language models generate convincing false content. The Chinese government conducted investigations and brought charges against offenders.
What happened
AI-generated misinformation spread widely in China in 2025, ranging from fabricated videos of landslides and nuclear leaks to false news about deaths of people in their 80s, gas explosions, and disease outbreaks. One organization analyzed social media trending topics using AI and generated 4,000 to 7,000 fake news articles daily via approximately 800 accounts, with some articles reaching over 1 million views.
Some misinformation resulted from 'AI hallucination'—where large language models output false information that appears plausible but lacks factual basis—particularly when trained on insufficient data in specialized fields. Other misinformation was deliberately created by individuals seeking traffic, views, or engagement on social media platforms.
Reported impact
AI-generated misinformation caused public panic and social disorder. False reports about landslides, gas explosions, and disease outbreaks alarmed local residents. Misinformation about Guangdong's medical insurance fund deficit caused social panic and public misunderstanding of policy. False rescue narratives about a 2025 Sichuan landslide hampered emergency response operations.
AI-generated misinformation spread rapidly and was difficult to counter. Unlike traditional rumors, AI-created falsehoods could evade keyword-based detection systems by rewording key terms, and multiple AI-generated accounts could amplify and reinforce each other.
Index analysis
AI misinformation was facilitated by low barriers to creation—open-source tools could generate realistic false images in 30 seconds with a simple text prompt, replacing the need for technical skills like photo editing. The technology enabled customized, multi-platform distribution at low cost with potential for exponential spread.
AI-generated falsehoods were harder to distinguish from truth than traditional misinformation. They appeared logically coherent and were designed to bypass technical screening. Some intentionally removed AI generation markers and exploited emotional triggers like 'disaster plus children' to drive rapid sharing.
Governance controls
In April 2025, China's Central Cyberspace Administration deployed a special campaign to address AI technology misuse, focusing on misinformation and false information created using AI. Police arrested individuals involved in mass-scale misinformation operations and brought administrative charges against users who created and spread false content, including those who used AI to fabricate specific incidents.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.