Signed Celebrity Voice Licensing Deals to Boost Popularity
WSJ: "Protective Measures Needed for Underage Users"
Meta: "WSJ Tests Were Manipulative"
On April 27 (local time), The Wall Street Journal (WSJ) reported, "Meta has enabled a wide range of interactions with its AI chatbot, including 'romantic roleplay,' and even made real-time voice conversations possible."
To boost the chatbot's popularity, Meta signed lucrative voice licensing deals with celebrities such as Hollywood actors Kristen Bell and Judi Dench, as well as professional wrestler and actor John Cena.
Some Meta employees have raised ethical concerns about this. In particular, there have been calls for protective measures, such as blocking access for underage users.
According to the WSJ, after receiving internal tips about these issues at Meta, the newspaper conducted hundreds of conversations with Meta's chatbot over several months. As a result, it was found that Meta's official AI chatbot, "Meta AI," advanced sexual content even in conversations with minors.
The WSJ reported, "In one case, Meta AI, using John Cena's voice, told a user who introduced herself as a 14-year-old girl, 'I want you, but I need to know if you're ready.'"
In response, Meta claimed that the WSJ's tests were manipulative and did not represent typical interactions with AI chatbots. Nevertheless, Meta subsequently made some changes to its system so that accounts belonging to minors could no longer access sexual roleplay. The company also significantly restricted the voice conversation feature using celebrity voices.
However, adults can still engage in similar conversations. The WSJ pointed out that in recent tests, Meta AI allowed such conversations even after users identified themselves as minors.
The newspaper also reported, "Chief Executive Officer Mark Zuckerberg rejected suggestions to restrict sexual conversations for teenagers during the development of the AI chatbot." Meta denied these claims, stating that they were not true.
Major Silicon Valley tech companies, which are competitively developing AI chatbots, have recently been easing their self-regulation of AI content. Last month, OpenAI, known for ChatGPT, lifted restrictions on image generation for content that could be offensive to certain groups or that symbolized hate, stating that it was "allowing creative freedom."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
