Lawyers for Scarlett Johansson are demanding that OpenAI disclose how it developed an AI personal assistant voice that the actress says sounds uncannily similar to her own.
Johansson’s legal team has sent OpenAI two letters asking the company to detail the process by which it developed a voice the tech company dubbed “Sky,” Johansson’s publicist told NPR in a revelation that has not been previously reported.
↫ Bobby Allyn at NPR
This story highlights just how much disdain techbros have for the work of creative people. Here’s the timeline:
- Nine months ago, Sam Altman approached Scarlett Johansson to ask her if OpenAI could use her voice for a voice assistant features. Johansson declined.
- Two days before the launch of the new voice assistant feature, Altman contacted Johansson’s agent again, asking her to reconsider.
- Before Johansson or her agent could reply, OpenAI launched the voice assistant, with a voice that sounds remarkably like Johansson’s. Altman even tweeted “Her”, the name of the film in which Johansson portrays an AI.
- After everyone started pointing this out, Johansson’s lawyers demanded OpenAI take the new voice down. They complied.
Techbros like Sam Altman deeply despise and undervalue the work of creatives, believing human creativity to be merely an equation to be solved, definable by an algorithm. To people like him, creative work has no value, and as such, is up for grabs to be taken and cut up for his algorithms to spit out as “new” works. This story highlights this perfectly.
The sleaze runs deep with Altman and OpenAI.
Thom Holwerda,
I’ll try not to take this too personally 🙂
I’m noticing a bit of a shift in the narrative around AI. Not so long ago it used to be that AI was considered too dumb to replace working people but I always felt this was naive. I think it’s a good thing that we’re getting over the “can AI replace us” denial and moving towards “should AI replace us”. It’s only going to become more capable and specialized over the next decade. Right now we are in the critical moments that will shape the future. This is not altogether different from the early days of computing that would shape computers and the internet for decades to come. What is AI’s role in society and how do we make sure it serves the good of the public at large? Decades from now what are the future regrets going to be? “We wouldn’t be in the mess if only they did it right at the beginning”. We are at the beginning and if we pull the right levers today, it may be possible to change norms, standards, and expectations before it’s too late and the norms become set without our input.
When you speak of norms, Johansson said no and Altman still went and did it. As for glorifying this as being something new, or special, due to AI buzzword being used. You have to admit that is kinda dumb. Pun intended.
Geck,
I agree. They did the cardinal sin of asking first. Companies should never ask first, unless they are ready for a “no” answer.
I am half joking of course.
If they really used her voice to train the AI, they should have asked.
But if the AI happened to sound like her artificially, and they just wanted to get her blessing, they should not have asked.
Alfman.
It seems like we are going through a version of “5 stages of grief”. Just that some of us do it faster than others, while many are still in denial phase. It will get better of course with never generations that never needing this debate.
While I do not support OpenAI in making a voice like this, especially for its suductive quality, calling people “techbros” and considering one group of intellectual workers creatives, while shutting on another is not cool. Consider AI, as an example. People made a machine talk like a human. It may not be a bright mind, but it is a mind nonetheless you are talking with there. How is creating that not creative, yet talking into a microphone, like actors do, is?
I don’t really disagree with your point, but I do disagree with your assessment of actors. You are being as condescending to them as you are accusing Thom of being towards Altman et al. Actors do much more than “talk into a microphone”, they act with their entire body and mind and voice. Like any other creative talent, it is part natural charisma and part learned skill, and it is hard work to maintain one’s craft.
Of course, I agree. I just wanted to demonstrate how easy it is to trivialize both ways.
In the days prior to release they asked permission. They didn’t get it. As soon as she complained the changed the voice. These people knew they were sailing close to the wind and hoped they’d get away with it.
The absolute gall of OpenAI for they to think they can just copy everything and get away with it. More and more, they are truly becoming a Massive Plagiarism Machine
In a news clip covering this topic, I just heard that the voice actor behind apple siri apparently didn’t realize her voice was being used by apple.
https://www.businessinsider.com/original-voice-of-siri-voice-actor-apple-used-her-voice-2023-2
Alfman,
This is actually quite common in the corporate world.
I had seen upcoming product domains being purchased by random third party intermediaries. This not only keeps the price reasonable (“Hmm… someone is asking my my uniquename.com for $8,000” vs “Billion dollar company is asking my domain, the price is $100,000 minimum”) but it also hides product surprises.
Even Disneyland famously bought the land in parcels in Florida when building their own castle / city.
One might think this is not fair. But they are paying fair market rates, and usually a bit above that.
(The only concern would be if the was for something you’d otherwise object on ethical reasons. Like using your photo for “before” in personal improvement products, and so on).
sukru,
I’m not sure if we can call it a fair market rate if one party has the power to keep the other party in the dark over the nature of the transaction.
I think of a fair market price as something none of the parties regrets afterwards and would do again. If there is deception involved then it’s probably a fishy deal. Obviously I cannot say it doesn’t happen, only that I disagree with the notion that deception doesn’t create an ethical issue in and of itself.
Alfman,
You have a valid point, and this is a bit nuanced.
But I could give the “eminent domain” as an example. They would (hopefully) give the fair market price for the property regardless of the reason. Though, to be fair again, this has been abused by cronies for taking over land for commercial development with the heavy hand of government.
Again, I would still object to using the results in a context you would not approve.
Shouldn’t happen anymore. One of the last protests from the screen actors guild got voice actors royalties instead of just pay per a line roles. Mainly due to South Park piecing together sound clips of Chef for an episode after he quit the show, and also the Dragonball voice actor that permanently destroyed his voice. But in 2005, yeah, pay per line. Not much say in what happens to those recordings later.
Update on the story:
https://www.washingtonpost.com/technology/2024/05/22/openai-scarlett-johansson-chatgpt-ai-voice/
This is a third option I did not think about. They asked her, she declined, and they found someone else with similar voice, and called the final work “Sky”.
If there indeed is another human, I don’t think there is any more case here.
sukru,
People’s voices and looks aren’t that unique given a large enough data set.
The “birthday attack” isn’t usually applied in this context, although I think it’s perfectly relevant.
Exactly. Your voice isn’t as unique as you think it is. Thom’s resentment towards AI (for taking his job) is so severe that he is essentially proposing a massive expansion of copyright that would make 2014 Thom Holwerda really hate 2024 Thom Holwerda.
Just imagine if copyright worked this way: “You hired an actress with a voice similar to mine, you owe me a license!”