Elections around the world face an evolving threat from foreign actors, involving artificial intelligence.
Countries trying to influence each other’s elections entered a new era in 2016, when the Russians launched a series of social media disinformation campaigns targeting the US presidential election. Over the next seven years, several countries – most notably China and Iran – used social media to influence foreign elections, both in the US and elsewhere around the world. There is no reason to expect 2023 and 2024 to be any different.
But there is a new element: generative AI and large language models. These have the ability to quickly and easily produce endless reams of text on any topic in any tone and from any perspective. As a security expert, I believe this is a tool particularly suited to propaganda in the internet age.
This is all very new. ChatGPT was introduced in November 2022. The more powerful GPT-4 was released in March 2023. Other language- and image-producing AIs are about the same age. It’s unclear how these technologies will change disinformation, how effective they will be, or what effects they will have. But we’re about to find out.
Election season will soon be in full swing in much of the democratic world. Seventy-one percent of people living in democracies will vote in national elections between now and the end of next year. These include Argentina and Poland in October, Taiwan in January, Indonesia in February, India in April, the European Union and Mexico in June, and the United States in November. Nine African democracies, including South Africa, will have elections in 2024. Australia and the United Kingdom don’t have set dates, but elections are likely to be held in 2024.
Many of these elections matter a lot to countries that have conducted social media influence operations in the past. China cares a lot about Taiwan, Indonesia, India, and many African countries. Russia cares about the UK, Poland, Germany, and the EU in general. Everyone cares about the US.
And that’s just the biggest players. Every US national election since 2016 has brought with it an additional country attempting to influence the outcome. First it was just Russia, then Russia and China, and more recently those two plus Iran. As the financial cost of foreign influence decreases, more countries can get in on the act. Tools like ChatGPT significantly reduce the price of producing and distributing propaganda, bringing that capability into the budgets of many more countries.
A couple of months ago I attended a conference with representatives from all the US cybersecurity agencies. They talked about their expectations for election interference in 2024. They expected the usual players – Russia, China, and Iran – and one new and significant one: “domestic actors.” This is a direct result of this cost reduction.
Of course, running a disinformation campaign requires more than just generating content. The hard part is distribution. A propagandist needs a pool of fake accounts to post content on, and others to spread it into the mainstream where it can go viral. Companies like Meta have gotten much better at identifying these accounts and removing them.
Just last month, Meta announced that it had removed 7,704 Facebook accounts, 954 Facebook pages, 15 Facebook groups, and 15 Instagram accounts associated with a Chinese influence campaign, and identified hundreds more accounts on TikTok, X (formerly Twitter), LiveJournal, and Blogspot. But that was a campaign that began four years ago, producing pre-AI disinformation.
Disinformation is an arms race. Both attackers and defenders have gotten better, but so has the world of social media. Four years ago, Twitter was a direct line to the media, and propaganda on that platform was a way to upend the political narrative. A Columbia Journalism Review study found that most major news outlets used Russian tweets as sources of biased opinion. That Twitter, with virtually every news editor reading it and everyone posting there, is no more.
Many propaganda outlets have moved from Facebook to messaging platforms like Telegram and WhatsApp, making them harder to identify and remove. TikTok is a newer, Chinese-controlled platform better suited to short, provocative videos, which AI makes much easier to produce. And the current crop of generative AI is being connected to tools that will make content distribution easier, too.
Generative AI tools also enable new production and distribution techniques, such as low-level propaganda at scale. Imagine a new AI-powered personal account on social media. For the most part, it behaves normally. It posts about its fake daily life, joins interest groups, comments on other people’s posts, and generally behaves like a normal user. And every once in a while, not very often, it says—or amplifies—something political. These persona bots, as computer scientist Latanya Sweeney calls them, have negligible influence on their own. But replicated by the thousands or millions, they would have a lot more.
This is just one scenario. It’s likely that the military officials in Russia, China, and elsewhere responsible for election interference have their best people thinking about others. And their tactics will likely be much more sophisticated than they were in 2016.
Countries like Russia and China have a history of testing both cyberattacks and information operations on smaller countries before deploying them on a large scale. When this happens, it’s important to be able to detect these tactics. Countering new disinformation campaigns requires the ability to recognize them, and recognizing them requires seeking them out and cataloging them now.
In the cybersecurity world, researchers recognize that sharing attack methods and their effectiveness is the only way to build strong defenses. The same reasoning applies to these information campaigns: the more researchers learn about what techniques are being used in faraway countries, the better they can defend their own countries.
Disinformation campaigns in the AI era are likely to be much more sophisticated than they were in 2016. I think the United States needs to work hard to detect and identify AI propaganda in Taiwan, where a presidential candidate claims a deepfake audio recording was used to defame him, and elsewhere. Otherwise, we won’t see them when they get here. Unfortunately, researchers are being targeted and harassed instead.
Maybe everything will be okay. There have been some important democratic elections in the age of generative AI without significant disinformation problems: primaries in Argentina, first-round elections in Ecuador, and national elections in Thailand, Turkey, Spain, and Greece. But the sooner we know what to expect, the better we can deal with what’s to come.
Elections around the world face an evolving threat from foreign actors, involving artificial intelligence.
Countries trying to influence each other’s elections entered a new era in 2016, when the Russians launched a series of social media disinformation campaigns targeting the US presidential election. Over the next seven years, several countries – most notably China and Iran – used social media to influence foreign elections, both in the US and elsewhere around the world. There is no reason to expect 2023 and 2024 to be any different.
But there is a new element: generative AI and large language models. These have the ability to quickly and easily produce endless reams of text on any topic in any tone and from any perspective. As a security expert, I believe this is a tool particularly suited to propaganda in the internet age.
This is all very new. ChatGPT was introduced in November 2022. The more powerful GPT-4 was released in March 2023. Other language- and image-producing AIs are about the same age. It’s unclear how these technologies will change disinformation, how effective they will be, or what effects they will have. But we’re about to find out.
Election season will soon be in full swing in much of the democratic world. Seventy-one percent of people living in democracies will vote in national elections between now and the end of next year. These include Argentina and Poland in October, Taiwan in January, Indonesia in February, India in April, the European Union and Mexico in June, and the United States in November. Nine African democracies, including South Africa, will have elections in 2024. Australia and the United Kingdom don’t have set dates, but elections are likely to be held in 2024.
Many of these elections matter a lot to countries that have conducted social media influence operations in the past. China cares a lot about Taiwan, Indonesia, India, and many African countries. Russia cares about the UK, Poland, Germany, and the EU in general. Everyone cares about the US.
And that’s just the biggest players. Every US national election since 2016 has brought with it an additional country attempting to influence the outcome. First it was just Russia, then Russia and China, and more recently those two plus Iran. As the financial cost of foreign influence decreases, more countries can get in on the act. Tools like ChatGPT significantly reduce the price of producing and distributing propaganda, bringing that capability into the budgets of many more countries.
A couple of months ago I attended a conference with representatives from all the US cybersecurity agencies. They talked about their expectations for election interference in 2024. They expected the usual players – Russia, China, and Iran – and one new and significant one: “domestic actors.” This is a direct result of this cost reduction.
Of course, running a disinformation campaign requires more than just generating content. The hard part is distribution. A propagandist needs a pool of fake accounts to post content on, and others to spread it into the mainstream where it can go viral. Companies like Meta have gotten much better at identifying these accounts and removing them.
Just last month, Meta announced that it had removed 7,704 Facebook accounts, 954 Facebook pages, 15 Facebook groups, and 15 Instagram accounts associated with a Chinese influence campaign, and identified hundreds more accounts on TikTok, X (formerly Twitter), LiveJournal, and Blogspot. But that was a campaign that began four years ago, producing pre-AI disinformation.
Disinformation is an arms race. Both attackers and defenders have gotten better, but so has the world of social media. Four years ago, Twitter was a direct line to the media, and propaganda on that platform was a way to upend the political narrative. A Columbia Journalism Review study found that most major news outlets used Russian tweets as sources of biased opinion. That Twitter, with virtually every news editor reading it and everyone posting there, is no more.
Many propaganda outlets have moved from Facebook to messaging platforms like Telegram and WhatsApp, making them harder to identify and remove. TikTok is a newer, Chinese-controlled platform better suited to short, provocative videos, which AI makes much easier to produce. And the current crop of generative AI is being connected to tools that will make content distribution easier, too.
Generative AI tools also enable new production and distribution techniques, such as low-level propaganda at scale. Imagine a new AI-powered personal account on social media. For the most part, it behaves normally. It posts about its fake daily life, joins interest groups, comments on other people’s posts, and generally behaves like a normal user. And every once in a while, not very often, it says—or amplifies—something political. These persona bots, as computer scientist Latanya Sweeney calls them, have negligible influence on their own. But replicated by the thousands or millions, they would have a lot more.
This is just one scenario. It’s likely that the military officials in Russia, China, and elsewhere responsible for election interference have their best people thinking about others. And their tactics will likely be much more sophisticated than they were in 2016.
Countries like Russia and China have a history of testing both cyberattacks and information operations on smaller countries before deploying them on a large scale. When this happens, it’s important to be able to detect these tactics. Countering new disinformation campaigns requires the ability to recognize them, and recognizing them requires seeking them out and cataloging them now.
In the cybersecurity world, researchers recognize that sharing attack methods and their effectiveness is the only way to build strong defenses. The same reasoning applies to these information campaigns: the more researchers learn about what techniques are being used in faraway countries, the better they can defend their own countries.
Disinformation campaigns in the AI era are likely to be much more sophisticated than they were in 2016. I think the United States needs to work hard to detect and identify AI propaganda in Taiwan, where a presidential candidate claims a deepfake audio recording was used to defame him, and elsewhere. Otherwise, we won’t see them when they get here. Unfortunately, researchers are being targeted and harassed instead.
Maybe everything will be okay. There have been some important democratic elections in the age of generative AI without significant disinformation problems: primaries in Argentina, first-round elections in Ecuador, and national elections in Thailand, Turkey, Spain, and Greece. But the sooner we know what to expect, the better we can deal with what’s to come.