Microsoft issues a warning on Chinese AI meddling in elections

Beijing conducted a trial in Taiwan using AI-generated content to sway voters from a pro-sovereignty candidate

Microsoft has warned that China is preparing to use AI-generated content to rig elections in the US, South Korea, and India after a trial run during the Taiwanese presidential election. The internet giant believes that major elections in 2024 would be targeted by cyber groups funded by the Chinese state, perhaps with assistance from North Korea. The threat intelligence team at Microsoft said, “As populations in India, South Korea, and the United States head to the polls, we are likely to see Chinese cyber and influence actors, and to some extent North Korean cyber actors, work towards targeting these elections.”

Microsoft stated that China will, at the very least, produce and share AI-generated content on social media to promote their stances in these important elections. While the current impact of AI-generated content is deemed small, Microsoft cautioned that this could change. “While the impact of such content in swaying audiences remains low, China’s increasing experimentation in augmenting memes, videos, and audio will continue – and may prove effective down the line,” Microsoft noted.

Microsoft noted in the study that China has previously attempted to influence the results of the January presidential election in Taiwan using artificial intelligence. According to the business, this was the first time that an organisation with state support had used artificial intelligence (AI)-generated information to sway a foreign election.

A Beijing-backed organisation called Storm 1376, also known as Spamouflage, Dragonbridge, or Spamouflage, was very active during the Taiwanese election. Its attempts to sway the election included uploading a phoney audio tape to YouTube. Although Terry Gou withdrew from the contest in November, the film appeared to show him backing a different candidate. According to Microsoft, the video was “likely AI generated.” The video was taken down by YouTube before it could get really popular.

The Beijing-supported group disseminated a series of AI-generated memes targeting William Lai, the ultimately successful candidate in Taiwan’s election. Lai, a pro-sovereignty candidate opposed by Beijing, faced baseless accusations of embezzling state funds in these memes. Additionally, there was an uptick in the use of AI-generated TV news anchors, a tactic also employed by Iran. These “anchors” made unfounded claims about Lai’s personal life, including allegations of fathering illegitimate children.

Microsoft noted that these news anchors were created using the CapCut tool, developed by Chinese company ByteDance, which also owns TikTok.

Furthermore, Microsoft highlighted that Chinese groups are continuing to conduct influence campaigns in the US. They are utilizing social media accounts to pose “divisive questions” and try to understand the issues that divide American voters.

In a blog post accompanying the report, Microsoft stated, “This could be aimed at gathering intelligence and insights into crucial voting demographics ahead of the US Presidential election.”

In one post on X, formerly Twitter, there was mention of a $118 billion bipartisan bill in the US. This bill included a $20 billion investment in the US-Mexico border along with a $75 billion package for Ukraine and Israel. The post asked for reactions to this news. Another post highlighted the loss of an F-35 fighter in South Carolina last year, stating that such an incident could only happen under the Biden administration. However, debris from the fighter was found soon after. The post asked for opinions on this event.

The report was released in the same week as a White House-appointed official review board stated that “a cascade of errors” by Microsoft had allowed state-backed Chinese cyber operators to access email accounts of senior US officials. Last month, both the US and UK governments accused China-backed hackers of conducting a prolonged cyber campaign targeting politicians, journalists, businesses, and even the UK’s election watchdog.