OpenAI, the creators of AI assistant ChatGPT, revealed yesterday that they acted within 24 hours last month to stop an online influence campaign aimed at India’s elections. The campaign, code-named “Zero Zeno”, was operated by an Israeli political consultancy called STOIC and sought to generate social media posts, comments and articles that criticized the ruling BJP party and praised the opposition Congress party.
Using sophisticated language models, the actors behind Zero Zeno tried to produce large volumes of content across platforms such as Facebook and YouTube with the goal of swaying Indian voters. However, OpenAI’s internal investigations soon detected these attempts at using their technology for covert and deceptive ends. They disrupted the related online activity and banned accounts connected to STOIC just one day after the campaign targeting India began in early May.
OpenAI disclosed that Zero Zeno was also targeting audiences in North America and Israel with English and Hebrew language content prior to expanding its focus to India and English. Their disruption of this operation marks the fifth such covert influence effort uncovered over the past three months that aimed to abuse their models. OpenAI remains committed to safeguarding the responsible development and application of AI.
In reaction, an Indian government minister warned that influence operations like this pose a serious threat to democracy in India. They emphasized the need for thorough investigation since it appears domestic political forces may have been involved in driving the foreign interference as well. As India’s national elections near their conclusion, OpenAI’s timely intervention has helped ensure voters can make informed choices without undue outside manipulation.
