Voices: What ChatGPT has done to Scarlett Johansson should scare us all
OpenAI’s much-criticised decision to use a Scarlett Johansson-like voice for its new chatbot, despite the actress’s refusal of requests to use her likeness, has shed fresh light on some of the industry’s murky practices when it comes to working with the creative community. AI has become a tool that corporations use to exploit artists and creators without remuneration or consent. While Johansson has pushed back against OpenAI, most creatives have far less power to negotiate or object.
The actress has since released a statement claiming that OpenAI CEO Sam Altman was pleading with her for months to use her voice, and even put in a final request two days before the launch. She refused, but OpenAI released the Johansson-like voice “Sky” anyway.
OpenAI has since attempted to backpedal, claiming “Sky’s voice is not an imitation of Scarlett Johansson” and that “AI voices should not deliberately mimic a celebrity’s distinctive voice”. Did Sam Altman not tweet “her” on the day of the OpenAI demo? Is Her not his favourite movie? Was the near-universal reaction of commentators not that it sounded eerily like the actress?
The creepy gender dynamics of the whole controversy are striking. It speaks volumes of an industry that has been systemically hostile to women and has a tendency to “move fast”, sometimes without consent.
OpenAI has since removed the voice, but the broader concern is that not all creatives have the fame and power of Johansson to defend their rights. In 2023, Irish voice actor Remi Michelle Clarke signed a contract with Microsoft to voice Bing, only later to discover a synthetic version of her voice marketed on another company’s AI voice website. What she didn’t realise at the time was that the contract allowed third parties to use her recordings in perpetuity. Her lawyers informed her that while the contract might be unethical, she had limited legal avenues to prevent the company from using a likeness of her voice.
What each of these incidents demonstrates is a potentially exploitative relationship between tech companies and creatives. Artists are justifiably concerned that AI companies are selling tools that produce knock-offs of their work without remuneration or credit. For example, David Holz, the founder of AI software Midjourney, admitted to using living artists’ work still under copyright without their consent to train the company’s AI-powered image generator.
These actions reflect a mindset where human artists are simply raw material turned into inputs for a voracious machine that feeds off human creativity and labour. This is also reminiscent of Apple’s recent infamous ad for its new iPad Pro that used a giant hydraulic press to crush creative instruments into its chip. This tone-deaf messaging expresses how many tech executives picture human creativity: something to be squeezed to produce their latest gadgets.
Copyright law that dates back centuries is struggling to keep up with advances in this technology. Generative AI companies have asserted they are legally entitled to use copyrighted works to train their software under a “fair use” exception, a point that has been vigorously contested by creatives and publishers in the courts.
Whatever the outcome of these cases, it is important to understand that this exploitative relationship between corporations and artists did not begin with AI. As part of research for our book Feeding the Machine: the Hidden Human Labour Powering AI, we spoke with numerous creatives whose lives had been transformed by AI. They pointed to a litany of problems ranging from uneven power dynamics to unfair terms in their contracts and being asked to sign away their rights to future revenue from their work. As one voice actor told us: “The issue of the creator being the least important person in the value chain and everyone else being on top of them has been around forever. AI hasn’t created the problem of exploitation, it has just exacerbated it.”
The problem isn’t so much with the technology but how it is being used by corporations to increase their profits and further exploit artists. Creatives have always faced people making reproductions of their works, but AI poses a completely new threat at a scale never before seen. Creative workers are fighting back, but governments must step in to enact new legislation that protects the rights of the creative community.
Dr James Muldoon is a reader in management at the University of Essex and author (with Mark Graham and Callum Cant) of ‘Feeding the Machine: the Hidden Human Labour Powering AI’