Propagandists seeking to influence elections around the world have attempted to use ChatGPT in their operations, according to a report released Wednesday by the technology’s maker, OpenAI.
While ChatGPT is generally considered one of the leading AI chatbots on the market, it also heavily moderates how people use the product. OpenAI is the only major tech company to repeatedly publish public reports on how malicious actors have attempted to abuse its Large Language Model, or LLM, product, providing some insight into how propagandists and criminal or state-sponsored hackers have attempted to leverage the technology and potentially use it with other AI models.
OpenAI said in its report that this year it has stopped people who tried to use ChatGPT to generate content about elections in the U.S., Rwanda, India and the European Union. It’s not clear whether many people were watching.
In one case, the company described an Iranian propaganda operation of fake English-language news websites that claimed to reflect various U.S. political views, though it’s not clear whether those sites ever received substantial engagement from real people. They also used ChatGPT to create social media posts supporting those sites, the report said.
In a media interview last month, U.S. intelligence officials said that propagandists working for Iran, as well as Russia and China, have all incorporated AI into their ongoing propaganda operations targeting American voters, but that none appear to have had any real success.
Last month, the U.S. charged three Iranian hackers who it said were behind an ongoing operation to hack and release Donald Trump’s presidential campaign documents.
Another operation that OpenAI says is linked to people in Rwanda was used to create biased posts on X in favor of the Patriotic Front, the repressive party that has ruled Rwanda since the end of the country’s genocide in the early 1990s. They were part of a larger campaign that repeatedly spammed pro-party posts on X, a documented propaganda campaign that posted messages — often the same few messages — more than 650,000 times.
The company also blocked two campaigns this year — one creating social media comments about the EU parliamentary elections and another creating content about the Indian general election — very soon after they began. Neither received any substantial engagement, OpenAI said, but it’s also unclear whether the people behind the campaigns simply moved on to other AI models created by other companies.
OpenAI also described how a specific Iranian hacking group targeting water and wastewater treatment plants repeatedly attempted to use ChatGPT in multiple stages of its operation.
A spokesperson for Iran’s mission to the United Nations did not respond to an email requesting comment on the water treatment plant hacking campaign or propaganda operation.