Google Thwarts Over 10,000 Attempts by Chinese Influence Operator

Written by

Google blocked over 10,000 instances of Dragon Bridge activity in Q1 2024, a China-affiliated influence operator that pushes pro-People’s Republic of China (PRC) views online.

The tech giant also disrupted over 65,000 instances of Dragon Bridge activity across YouTube and Blogger in 2023.

The Threat Analysis Group (TAG) at Google described the group as the most prolific influence operator it tracks.

The majority of the group’s activity remains low quality content without a political message, however, a small fraction of Dragon Bridge accounts post about current events, including elections in Taiwan and the Israel-Hamas war. Much of this content includes content critical of the US government.

The content primarily targets Chinese speakers, but some narratives are in English and other languages.

Growing Use of AI Tools in Influence Ops

Google observed a surge of Dragon Bridge activity around Taiwan’s general election, held on January 13, 2024.

This was the largest observed Dragon Bridge campaign that has used generative AI tools to create content to date, according to Google. In the days leading up to the election, the group posted thousands of videos and comments on YouTube using synthetic audio and avatars promoting a false “secret history” document critical of the outgoing Taiwan President Tsai Ing-wen.

The same narrative was pushed on multiple non-Google platforms including X (formerly Twitter), Reddit, Instagram and Facebook.

The influence operator also produced video content featuring AI-generated news hosts speaking to the camera in the style of a news broadcast.

Google said the quality of these videos varied, with some appearing realistic at first glance and others obviously computer animated.

The campaign around Taiwan’s election follows several years of Dragon Bridge experimenting with AI tools, dating back to 2022. This includes a Mandiant report in March 2023 describing how the group leveraged several AI-generated images to support narratives negatively portraying US leaders.

Dragon Bridge’s use of AI-generated videos or images has not resulted in significantly higher engagement from real viewers so far, Google noted.

US-Focused Content in Lead Up to November Elections

Dragon Bridge has continued to create content critical of US government, society and democracy in recent months, Google’s TAG found.

In May 2024, it created videos and comments in English commenting on the student protests over the Israel-Hamas war on US university campuses.

The content narrative was generally pro-Palestine, framing US and Western governments and media as hypocritical – for example, contrasting the police detaining protesting students in the US with the Western media’s support of student protests in Hong Kong amid growing Chinese influence in the region.

The content relating to the student protests was typical of previous Dragon Bridge campaigns according to Google’s Threat Analysis (TAG). The content included a mix of low-quality videos made of repurposed footage with text overlaid on top, and content packaged to look similar to a TV news story, with opening graphics, dramatic music and a news anchor.

Google believes it is highly likely that Dragon Bridge will develop pre-produced content relating the 2024 US general election.

In previous US elections, Dragon Bridge has not explicitly advocated for specific candidates or outcome. For example, in the 2020 Presidential election, its content was critical of both candidates. In the 2022 US midterm elections, Dragon Bridge attempted to spread narratives highlighting US political divisions, potential for political violence and threats to democracy.

Read here: US Election Officials Told to Prepare for Nation-State Influence Campaigns

TAG said that a portion of Dragon Bridge’s content on US politics on other platforms leveraged more nuanced tactics than had been previously observed. This includes using inauthentic personas posing as US residents and employing engagement strategies like "follow trains" to attract unwitting followers.

What’s hot on Infosecurity Magazine?