OpenAI’s report stated that its models are being used to influence elections. It also stated that it had taken down over 20 operations that relied on its AI model to carry out such malicious activities.
The OpenAI report , “An update on disrupting deceptive uses of AI,” also emphasized the need for vigilance when engaging with political content.
The document showed a trend with OpenAI’s models becoming a major tool for disrupting elections and spreading political misinformation. Bad actors, who are often state-sponsored, use these AI models for various activities, including generating content for fake personas on social media and malware reverse engineering.
OpenAI’s growing influence in AI elections and politics
In late August, OpenAI disrupted an Iranian campaign that was producing social media content to sway opinions in US elections, Venezuelan politics, the Gaza conflict, and Israel. It reported that some accounts, which were subsequently banned, were also posting about Rwandan elections.
It also found that an Israeli company was also involved in trying to manipulate poll results in India.
However, OpenAI noted that these activities have not gone viral or cultivated substantial audiences. Social media posts related to these campaigns gained minimal traction. This could indicate the difficulty in swaying public opinion through AI-powered misinformation campaigns.
Historically, political campaigns are usually fueled by misinformation from the running sides. However, the advent of AI presents a different threat to the integrity of political systems. The World Economic Forum (WEF) stated that 2024 is a historic year for elections, with 50 countries having elections.
LLMs in everyday use already have the capability to create and spread misinformation faster and more convincingly.
Regulation and collaborative efforts
In response to this potential threat, OpenAI said it is working with relevant stakeholders by sharing threat intelligence. It expects this collaborative approach to be sufficient in policing misinformation channels and fostering ethical AI use, especially in political contexts.
OpenAI reports, “Notwithstanding the lack of meaningful audience engagement resulting from this operation, we take seriously any efforts to use our services in foreign influence operations.”
The AI firm also stressed that robust security defenses must be built to prevent state-sponsored cyber attackers, who use AI to create deceptive and disruptive online campaigns.
The WEF has also highlighted the need to put AI regulations in place, saying, “International agreements on interoperable standards and baseline regulatory requirements will play an important part in enabling innovation and improving AI safety.”
Developing effective frameworks requires strategic partnerships between tech companies such as OpenAI, the public sector, and private stakeholders, which will help implement ethical AI systems.
