OpenAI has expressed concern that its new realistic AI voice feature might lead users to form emotional bonds with the technology at the expense of human connections. The San Francisco-based company highlighted research suggesting that engaging with AI in a human-like manner could result in misplaced trust, and the advanced quality of the GPT-4o voice could intensify this issue.

In a recent safety report on its ChatGPT-4o AI, OpenAI noted that users have begun interacting with the AI as if it were a real person, even expressing sentiments such as lamenting their “last day together.” While these interactions may seem harmless, OpenAI believes they warrant further investigation to understand their long-term effects.

The company also raised concerns that frequent socializing with AI might impair users’ skills in forming and maintaining human relationships. The report cautioned that AI’s ability to remember details and perform tasks could lead to over-reliance on the technology.

Alon Yamin, Co-Founder and CEO of AI anti-plagiarism platform Copyleaks, echoed these concerns, suggesting that it might be time to reassess how such technology impacts human interactions and relationships.

OpenAI plans to continue studying how its voice capabilities might influence emotional attachment. Additionally, recent incidents, such as the controversy over the use of a voice similar to actress Scarlett Johansson’s, have spotlighted the challenges and ethical considerations surrounding voice-cloning technology.