OpenAI Warns of Emotional Dependency Risks with ChatGPT’s Voice Feature

In a statement published on its website, OpenAI provided details about the safety studies conducted on the voice option offered for interacting with ChatGPT and the AI model. The initial investigations revealed that the voice option’s response time of 232 milliseconds is very close to the average human reaction time during conversation. The statement noted that users have been using language indicating they form an emotional bond with the technology due to its “realistic and natural” nature.

The statement also warned that the emotional connection established with the AI model’s voice feature could lead to “addiction.” While the “Voice Mode” might benefit individuals who are lonely, it could also impact human relationships. The statement further mentioned that the feature presents “new risks,” including unauthorized voice generation.

Additionally, the statement highlighted that the technology’s ability to complete tasks and remember details from conversations could foster “over-reliance and dependency” among users.