OpenAI, the leading artificial intelligence research and deployment company, has raised alarms about the exploitation of its AI models, ChatGPT and Dall-E, by foreign hacking groups with the intent to influence global elections. This announcement comes amidst a year touted as the most significant in terms of democratic participation, with over 50 countries set to engage in electoral processes.
The findings were detailed in a comprehensive 54-page report issued by OpenAI, highlighting 20 detected campaigns that have utilised AI to engage in electoral interference since the beginning of the year. This activity is expected to intensify as the United States approaches its presidential elections next month. Key concerns centre around the misuse of AI to fabricate articles, create fictitious personas, and engage with content on social media platforms, potentially swaying public opinion and disrupting democratic processes.
Sam Altman, CEO of OpenAI, previously voiced apprehensions regarding the implications of generative AI on election integrity. In his testimony before Congress last year, Altman expressed his unease about the unprecedented ways in which AI could facilitate the spread of misinformation.
The report indicates that the misuse of AI extends beyond basic disinformation tactics, encompassing sophisticated, multi-stage strategies designed to analyse and craft responses to social media interactions. OpenAI emphasised the crucial need for robust defences against state-linked cyber actors and covert influence operations. These operations exploit AI for deceptive campaigns across a broad spectrum of internet platforms.
The report specifically mentions hacking groups associated with state regimes in China, Iran, and Russia, identifying them as key players in these interference operations. A notable example cited was a "Russia-origin threat actor" who generated content in English and French, aimed at audiences in West Africa and the UK. This content included both short comments and long-form articles that were then disseminated through websites masquerading as legitimate news outlets.
Details regarding the targets of these campaigns reveal a unique approach to developing an audience base, with the UK-focused entities forming purported "information partnerships" with local organisations. However, these organisations, including a church in Yorkshire, a school in Wales, and an association of chambers of commerce in California, remain unnamed in the report. OpenAI's report refrains from divulging these specifics, prompting inquiries for further clarification.
As the deployment of artificial intelligence continues to evolve, OpenAI remains vigilant in monitoring its AI tools and implementing measures to thwart manipulative efforts. The company's ongoing efforts to intercept and dismantle deceptive networks underscore the broader challenges faced by technology firms in safeguarding democratic integrity in a digital age.
Source: Noah Wire Services