Chinese AI company leaked for cognitive warfare operations targeting Taiwan and 20+ countries
Leaked documents from Chinese company 安洵資訊 (iSoon) reportedly revealed a systematic campaign to spread false information and manipulate public opinion in Taiwan and over 20 other countries using AI-driven bot accounts and hacking tools provided to Chinese government agencies. The operations targeted vulnerable populations with misinformation about Indian workers, nuclear radiation, and Hong Kong protests, resulting in real-world protests and unrest.
What happened
Internal documents from Chinese company 安洵資訊 (iSoon) were leaked in February 2024, revealing the company provided data and hacking tools to Chinese intelligence and security agencies. The documents showed 577 files teaching website attack techniques, stolen product manufacturing data, and company communications spanning 3 years.
A document titled 'Twitter輿情導控' (Twitter Opinion Control) containing 21 pages of tactical guidance described how users clicking fake links would have accounts hijacked and converted into bot accounts used to spread false information. The system was developed specifically for Chinese public security agencies.
Documents revealed that hackers collected Taiwanese personal data including population data, road information, 3D building models of each county and city, 3D models and database information of national highways, and research from some universities. The data collection operations targeted over 20 countries and regions including Taiwan, Japan, Hong Kong, Europe and Africa.
Reported impact
In Taiwan, false information about importing 100,000 Indian workers circulated on the Dcard platform starting approximately 3 weeks before protests, falsely claiming Indian workers would increase sexual violence. Posts on social media amplified the narrative and prompted young female protesters to take to the streets; protesters reported learning about the issue from Dcard.
False information about Japanese nuclear contamination spread through bot accounts, claiming a 2011 radiation impact prediction map was showing impacts from released nuclear wastewater. The article was retweeted 2,300 times with 900,000 views.
During Hong Kong's 2019 anti-extradition movement, bot accounts criticized the protests and promoted pro-China messages. Affected countries reportedly experienced public unrest and demonstrations resulting from cognitive warfare operations.
Index analysis
The leaked documents showed the tool description explicitly stated that 'public security organs controlling opinion narratives is crucial for social stability' and that 'cognitive warfare is gradually entering public view and quietly changing the world order.' More than half of iSoon's transaction partners were Chinese public security agencies.
Documents indicated that iSoon employees discussed Taiwan National Chengchi University data as potentially useful for researching cross-strait relations. During the same period, the university's servers were attacked, suggesting use as a jumping-off point to access other institutions' information.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.