TAICHUNG (Taiwan News) — Several technologies will soon converge, unleashing massive torrents of high-quality, hard-to-spot propaganda and disinformation. Taiwan will be ground zero, and it will most likely happen in 2023, during the election campaign.

Taiwan already faces a large volume of propaganda and disinformation. The Chinese Communist Party’s (CCP) United Front Work Department (UFWD) has been actively targeting Taiwan for some time, and Taiwan is widely seen as ground zero and the testing ground for UFWD activities.

Financial Times reported during the 2020 national election cycle that reporters for both the Taiwanese news outlets Want Want China Times newspaper and CTiTV news received instructions directly from China’s Taiwan Affairs Office (TAO). These revelations led to abusive libel lawsuits against prominent journalist Kathrin Hille, though in the end they were fortunately dropped. CTiTV was eventually taken off cable TV, giving over half their entire news coverage to the then China-favored KMT presidential candidate Daniel Han Kuo-yu (韓國瑜), but China Times is still operating normally and CTiTV is now on Youtube.

Websites, social media pages, and content farms spreading Chinese propaganda are chronic problems. Hackers have taken over websites and social media accounts, and even briefly took over display screens at 7-11s and Taiwan Railway Administration stations.

Current methods are crude

There are armies of “50-centers” (aka “wumao”) who are said to be paid a half Renminbi for posts, and unpaid nationalistic “little pinks” that roam social media and comment sections spouting aggressive Chinese nationalism. Bot accounts on social media clumsily spread propaganda and together form the so-called “internet water army” that floods and overwhelms social media pages, such as that of Taiwan’s president and premier.

However, up to now, all these efforts have been fairly crude, highly labor-intensive, and usually easy to spot. Only a relatively small number of methods actually work, and usually only with people already inclined to the propaganda.

That being said, the sheer volume keeps praiseworthy civil society groups like Doublethink Lab and the recently created Ministry of Digital Affairs (MODA) very busy, and they are about to get significantly busier.

Artificial intelligence (AI) tools like ChatGPT are set to take off in China, with Baidu set to release their Chinese version in March. The UFWD will certainly get their hands on the technology, tweaked for their requirements.

This type of AI can write passable articles and even entire books with limited prompts very quickly, opening the floodgates to fill fake news sites and social media accounts with an unimaginable volume of content.

Previously, each person could write one or two articles per day. Now people will be able to pump out articles in a minute or less, with the only limitation being coming up with enough prompts and topics to feed the AI to keep the articles flowing.

Overwhelming volume

The tools will likely overwhelm any fact checkers, moderators, and other defenses. The sheer volume will also make it more likely that they will dominate search engine results for all sorts of keywords and phrases, crowd out legitimate journalism and academic papers, and clog results with propaganda to their liking.

This is concerning, as search engines are still our main go-to to find information. Worse, AI platforms like ChatGPT produce their results based on combing through the internet. While ChatGPT only has data up to 2021, Bing search is already testing its own without limitation, and ChatGPT will no doubt follow.

This could lead to both search engines and AI platforms being corrupted by UFWD propaganda and inadvertently doing their bidding. Both must do some serious thinking and planning to avoid this happening.

This is going to be linked with another AI development. Just recently, reporters at the Guardian, Le Monde, and Der Spiegel conducted an undercover investigation that found a service being offered out of Israel called Advanced Impact Media Solutions (Aims) that can manage over 30,000 social media accounts across many platforms.

These companies use avatars with a “multifaceted digital backstory” and the “avatars mimicked human behavior and their posts were powered by artificial intelligence.” In short, an AI-powered system is exponentially more powerful than the rather crude systems that create Twitter bots today, and it’s only a matter of time before the UFWD gets its hands on it, or something similar.

Realistic deepfake videos

AI is also going to create the ultimate monster, realistic deepfake videos that can, once fed a certain amount of video and audio (say from news clips) of a person, can recreate the image of that person and get that digital copy to say and do whatever they like. Deepfakes take a lot of work to create, and it has to be done by a person, but that’s about to change.

In just a few days, ChatGPT will be launching a video version. Of course, it will no doubt have plenty of safeguards against creating deepfakes, but the UFWD is almost certainly working on getting their hands on this technology.

Once they do, they can flood the internet with fake videos, having politicians or whomever they like either parroting their propaganda, or saying or doing something outrageous. It is not hard to see how this could be extremely damaging in a political campaign.

Most likely these tools will be used to undermine the Democratic Progressive Party (DPP) in order to get more China-friendly candidates elected. The DPP will need to think of a strategy to deal with this.