OpenAI has issued a stark warning about foreign hacking groups using its AI tools, ChatGPT and DALL-E, to interfere in elections. A newly released 54-page report details how these groups are attempting to manipulate democratic processes worldwide. Since the beginning of this year, OpenAI has detected 20 such campaigns, with more expected as the US presidential election approaches.
Hacking groups have used ChatGPT to write fake articles, create false personas, and post manipulative content on social media. These activities are part of sophisticated, multi-stage efforts to influence public opinion and interfere with democratic processes.
This year marks a significant moment for global democracy, with over 50 countries holding elections. However, the rise of generative AI has led to concerns about its misuse in election manipulation. OpenAI and other tech firms are stepping up efforts to combat potential election interference.
OpenAI’s chief executive, Sam Altman, has previously expressed concerns about AI’s potential to spread disinformation. He warned that generative AI could be used to influence elections in unprecedented ways.
The report points to hacking groups affiliated with regimes in China, Iran, and Russia as suspects in some of these interference operations. One case study highlights a Russia-based group that used OpenAI models to generate English and French content targeting West Africa and the UK, posting articles on fake news websites.
While specific organizations were not named, OpenAI continues to strengthen defences against these emerging threats to election integrity.
