The chatbot was developed by Indian AI developer Delphi AI Inc., using OpenAI’s GPT-4 model, and was contracted by the political action committee. Despite a disclaimer stating that responses were generated by a chatbot, not Phillips, OpenAI suspended the company's account and access to Dean.Bot was removed. While some see generative AI as a promising tool for voter education, there are concerns about misuse, including impersonation and disinformation. OpenAI has previously stated it does not allow applications for political campaigning and lobbying, including chatbots impersonating candidates.
Key takeaways:
- OpenAI has suspended a startup that developed a chatbot, Dean.Bot, which impersonates Democratic presidential candidate Dean Phillips. This is the first time OpenAI has censored a developer for violating its guidelines on AI misuse.
- The chatbot was developed by Silicon Valley entrepreneurs Matt Krisiloff and Jed Somers, and was used by a political action committee, We Deserve Better, to support Phillips ahead of the New Hampshire primary.
- The chatbot was powered by OpenAI’s most powerful large language model, GPT-4, and was developed by an Indian AI developer, Delphi AI Inc. The company's OpenAI account was suspended and access to Dean.Bot was removed.
- OpenAI has policies against the use of its technology for political campaigning and impersonating individuals without consent. Despite the potential of generative AI to educate voters, there are concerns about its misuse for impersonation and disinformation.