A Model's Fight Against AI-Generated Racial Alterations
Elii Emeghebo, a Nigerian-Australian model, recently found himself in a situation that left him deeply unsettled. Two years ago, while walking past Peter Jackson's Pitt Street shop in Sydney, he noticed an image in the window that looked eerily similar to him—except for one significant detail. The man in the image had features that were dramatically altered, appearing almost like his "white twin."
In an email to his agent, Mr. Emeghebo described the figure as his "white twin." He pointed out that his nose was reshaped, his skin tone and eye color were significantly lightened, and there were changes around his eyebrows and eye shape to make them more Eurocentric and less Black.
"It's not a good thing when you go from a shoot, something you're really proud of … and you walk past the shop and see basically you, but without your identity there," he said. "It was really confronting and really unfair."
Mr. Emeghebo alleges that the AI image "whitewashed" his features and believes he should have been paid for their use because he had only been contracted to shoot images for the brand's digital channels. He is now pursuing a complaint against Peter Jackson Australia at the Australian Human Rights Commission, alleging the unauthorized alterations to his image amounted to racial discrimination and offensive behavior.
Peter Jackson Australia admits using AI to alter Mr. Emeghebo's appearance but categorically denies discriminating against him based on his race. "Any suggestion that we intentionally engaged in racial discrimination is false, and we reject it unequivocally," a representative said in a statement. They explained that they had paid Mr. Emeghebo in full for the shoot and used his "unaltered images across multiple channels in line with our agreement." The representative added that the brand's creative team later used "AI-assisted tools" to produce a "substantially transformed image" from the original campaign featuring Mr. Emeghebo. "Any suggestion that this process was undertaken because of Mr Emeghebo's race, colour or ethnic origin is categorically denied," the statement said.
Victoria's Young Workers Centre assisted Mr. Emeghebo in preparing his discrimination complaint. Its principal legal officer, Paloma Buhagiar Cole, said it was hard to imagine an alternate reason for the changes to Mr. Emeghebo's image. "AI doesn't operate in a vacuum; it responds to prompts from human beings," she said. "Human beings chose to use those particular images and decided the image that is devoid of Elii's identity was the one they wanted to use."
Ms. Cole also alleged that Mr. Emeghebo was entitled to payment for the use of the AI-generated image in the shopfront. "He had a very specific contract that allowed for specific use … so they haven't actually paid for the use of that image here," she said.
Mr. Emeghebo wants to see changes to address the growing influence of artificial intelligence on his industry. "I'd like to see a bigger change that helps to prevent these sorts of issues from cropping up further with the development of AI, and affecting more models like myself and people of colour," he said.
AI Protections for Models 'Shoehorned' into Existing Law
Internationally, a number of models have taken action against the unauthorized use of AI to reproduce or alter their likeness. Popular New Zealand streetwear brand Huffer is facing backlash for using what the brand has described as an AI editing tool in their campaigns, as reported by ABC News Breakfast earlier this month. In New York, model Francheska Pujols has filed a lawsuit against fashion label Rainbow USA, alleging they used AI to generate synthetic images of her without her consent.
With AI use exploding worldwide, University of Melbourne academic and lawyer Jason Bosland said Australia should consider legislating a right for a person to control the use of their image, similar to those in place in several US states. Without an "image right," he said Australian models currently need to "shoehorn" AI provisions into existing legal protections. Those could include writing clauses about AI usage into models' contracts or pursuing "misleading or deceptive conduct" claims under consumer law.
"That would require that the model have a reputation in the industry and by using the model's image that creates a misrepresentation to the public that they have consented to or endorsed the brand," Mr. Bosland said.
No comments:
Post a Comment