It all started when Open AI’s CEO Sam Altman unveiled a new ChatGPT version that included a new voice assistant seemingly inspired by the movie Her. Altman had professed his love for the movie before, declaring it his favorite. Controversy started bubbling over how Scarlett Johansson’s AI assistant character influenced ChatGPT’s real-life AI voice assistant.  

Just like in Her, ChatGPT’s version can detect a person’s emotions by analyzing the user’s facial expressions and tone of voice to then, in turn, “emote” while responding to the user. Basically, it feels like you’re having an actual conversation with your assistant.  

One of the voices you could select to use was Sky, one that has been described as breathy and coquettish. (It even inspired a skit on The Daily Show to discuss how sexy it sounded.) However, trouble began when many noticed just how similar—perhaps too similar—it was to Scarlett Johansson’s real voice.  

For the actress, the issue began long ago, when OpenAI’s CEO personally reached out to convince her to lend them her voice. She declined. However, it didn’t stop the company from trying to replicate her whole vibe. When Johansson publicly denounced the company and got lawyers involved, OpenAI took down the Sky voice indefinitely. 

Why does this matter? 

AI has always been a topic of controversy. Artists and workers from many industries, including music, cinema, and animation, have denounced the technology, calling for more regulation and protections. Visual artists have started to fight back against the machine itself.  

Though this is technically not voice cloning, it does potentially use an artist’s brand and work to gain profit. So, how much inspiration is too much inspiration?  

Adding to that, this definitely confirms that any company owning such technology can mimic or clone another’s brand if they simply choose to. Scenarios like this one could help determine AI policy and applications moving forward.   

AI lobbying has increased significantly as calls for regulation keep coming in. However, the government is lagging behind AI’s advancements. There’s already a lot on the table and more will continue to be added.  

Issues in application 

OpenAI’s voice generation AI model just needs a 15-second sample to successfully generate a voice that closely resembles the original speaker—a truly impressive technological feat. There’s also a myriad of voice cloning apps out there. Unfortunately, in the era of fake news, scammers may use these tools to spread misinformation and threaten people’s digital safety.  

And it’s not only your run-of-the-mill scammer that has used it—others have begun to exploit it. For example, a political consultant who deepfaked Biden’s voice was indicted recently. And a Baltimore teacher was arrested after creating and spreading an audio deepfake of their principal making racists remarks. 

However, it’s important to note that the technology itself isn’t inherently bad. In fact, voice assisted technology could open the possibilities to people in need, like people with speech disorders. Yet, the technology is operating in an ethically murky territory at the moment.  

Keeping ourselves informed 

There are many questions about AI that remain open or undecided, yet there’s no denying that it’s a tool that will continue to take up space in our lives. We must continue to learn and keep an eye on new developments. Understanding the risks is key when protecting ourselves and our people while using new technologies.