OpenAI "Attempts to Influence Elections Using Generative AI Continue"

The Influence on the Public Is Revealed to Be Minimal

OpenAI revealed that attempts to influence elections worldwide using its generative AI, ChatGPT, continue. However, such 'election viral' activities have so far had a minimal impact on the public.

On the 9th (local time), OpenAI reported, "We have shut down more than 20 related networks worldwide that attempted to use our model." The types of election-related AI content varied, including web articles and fake social network service (SNS) account posts.

This OpenAI report was released less than a month before the U.S. presidential election. The U.S. election is considered a major event that could shake the global economic landscape. Although election misinformation is not new, this U.S. election has attracted significant attention regarding its potential impact on the outcome, as it is the first held since generative AI began to become widely popular.

OpenAI explained that most AI-generated election-related SNS content is related to elections in the U.S. and Rwanda, with little connection to elections in India or the European Union (EU).

However, the impact of such election-related SNS content on the public is still considered minimal.

OpenAI stated that at the end of August, an organization in Iran used ChatGPT to generate long-form articles and SNS comments about the U.S. election, but most posts did not receive any 'likes,' nor were there comments or shares.