OpenAI stated in January that it would prohibit the use of its technology for creating chatbots that imitate political candidates or disseminate false information related to voting. The company also declared that it would not allow the development of applications for political campaigns or lobbying.
Although the Kennedy chatbot page does not reveal the specific model behind it, the source code links the bot to LiveChatAI, a company known for providing GPT-4 and GPT-3.5-powered customer support chatbots to businesses. LiveChatAI’s website mentions that its bots utilize the capabilities of ChatGPT.
When questioned about the large language model used by the Kennedy campaign’s bot, LiveChatAI co-founder Emre Elbeyoglu stated that the platform incorporates technologies like Llama and Mistral in addition to GPT-3.5 and GPT-4. Elbeyoglu emphasized the platform’s commitment to client confidentiality, thus refraining from confirming or denying client specifics.
OpenAI spokesperson Niko Felix mentioned that there was no direct evidence of the Kennedy campaign chatbot leveraging OpenAI’s services, hinting at the possibility of LiveChatAI using one of its models through Microsoft’s services, given Microsoft’s significant investment in OpenAI since 2019.
Microsoft confirmed that the Kennedy chatbot utilizes the capabilities of Microsoft Azure OpenAI Service and clarified that their customers are not bound by OpenAI’s terms of service. They stated that the chatbot complies with Microsoft’s policies and is designed to prevent misinformation.
OpenAI did not respond immediately to queries about potential rule violations by the bot. Earlier this year, OpenAI blocked Dean.bot, a chatbot based on its models that imitated Democratic presidential candidate Dean Phillips.
The chatbot service was unavailable on Sunday afternoon but reappeared on Monday. The page displayed a red exclamation point icon indicating the chatbot was not found. Requests for comments from relevant parties regarding the chatbot’s removal went unanswered.
The use of chatbots in political contexts has sparked controversy due to their tendency to generate misleading information. While OpenAI explicitly prohibits electoral and campaigning use of its tools, other major language models like Meta, Microsoft, Google, and Mistral have terms of service that do not directly address political use.
Concerns have been raised about the enforcement of OpenAI’s policies given the accessibility of its tools through third parties for campaigns. The RFK Jr. campaign’s chatbot reappeared on Monday afternoon.
