Comment by HarHarVeryFunny
2 years ago
Clearly an SJ voice was the goal, given that Altman asked her to do it, asked her a second time just two days before the ChatGPT-4o release, and then tweeted "her" on the release day. The next day Karpathy, recently ex-OpenAI, then tweets "The killer app of LLMs is Scarlett Johansson".
Altman appears to be an habitual liar. Note his recent claim not to be aware of the non-disparagement and claw-back terms he had departing employees agree to. Are we supposed to believe that the company lawyer or head of HR did this without consulting (or more likely being instructed by) the co-founder and CEO?!
They hired the actor that did the voice months before they contacted SJ. The reaction on this site to the news that this story was false is kind of mindbending.
My guess: Sam wanted to imitate the voice from Her and became aware of Midler v. Ford cases so reached out to SJ. He probably didn't expect her decline. Anyway, this prior case tells that you cannot mimic other's voice without their permission and the overall timeline indicates OpenAI's "intention" of imitation. It does not matter if they used SJ's voice in the training set or not. Their intention matters.
Please don't take this as me defending OpenAI's clearly sketchy process. I'm writing this to help myself think through it.
If it weren't for their attempt to link the voice to SJ (i.e. with the "her" tweet), would that be OK?
- It's fine to hire a voice actor.
- It's fine to train a system to sound like that voice actor.
- It's fine to hire a voice actor who sounds like someone else.
- It's probably fine to go out of your way to hire a voice actor who sounds like someone else.
- It's probably not fine to hire a voice actor and tell them to imitate someone else.
- It's very likely not fine to market your AI as "sounds like Jane Doe, who sounds like SJ".
- It's definitely not fine to market your AI as "sounds like SJ".
Say I wanted to make my AI voice sound like Patrick Stewart. Surely it's OK to hire an English actor who sounds a lot like him, so long as I don't use Sir Patrick's name in the advertising. If so, would it have been OK for OpenAI to do all this as long as they didn't mention SJ? Or is SJ so clearly identifiable with her role in "Her" that it's never OK to try to make a product like "Her" that sounds like SJ?
32 replies →
A random person's normal speaking voice is nobody's intellectual property. The burden would have been on SJ to prove that the voice actor they hired was "impersonating" SJ. She was not: the Washington Post got her to record a voice sample to illustrate that she wasn't doing an impersonation.
Unless & until some 3rd other shoe drops, what we know now strongly --- overwhelmingly, really --- suggests that there was simply no story here. But we are all biased towards there being an interesting story behind everything, especially when it ratifies our casting of good guys and bad guys.
26 replies →
I cannot read the article because of it's paywall - is there actual proof OpenAI reached out to Johansson - or is it just being alleged by her lawyers?
It seems she has every reason to benefit from claiming Sky sounded like her even if it was a coincidence. "Go away" payments are very common, even for celebrities - and OpenAI has deep pockets...
Even so, if they got a voice actor to impersonate or sound similar to Johansson, is that something that's not allowed?
7 replies →
Sure, no-one is disputing that, and despite this Altman then contacts SJ again two days before release asking her to reconsider, then tweets "her" to remind the public what he was shooting for. The goal could have just been ChatGPT with a voice interface, but instead Altman himself is saying the the goal was specifically to copy "her".
He's not necessarily saying that was the goal from the start. All he is admitting with that tweet is that it is indeed (he finds it to be) reminiscent of "Her".
12 replies →
Lots of people are disputing it.
5 replies →
Even if the voice actor was sourced before they originally contacted SJ, it was clearly the intent to sound like her. There are so many other distinctive voices they could have chosen, but instead they decided to go as close as possible to "her" as they could. Many people thought it was SJ until she stated it wasn't. I appreciate the voice actor may sound like that naturally, but its hardly coincidental that the voice that sounds most like the voice from "her" was the one chosen for their promotion. It is clearly an attempt to pass-off.
>Even if the voice actor was sourced before they originally contacted SJ, it was clearly the intent to sound like her.
Her, being the voice SJ did for the movie, not SJ's conversational voice which is somewhat different.
If OpenAI were smart, they did it in a chinese wall manner and looked for someone whose voice sounded like the movie without involving SJ's voice in the discussion.
This is not a thing. They hired a voice actor, who spoke in her normal speaking voice. That voice is not SJ's intellectual property, no matter what it sounds like. Further, I don't know how you can say any intention here is "clear", especially given the track record on this particular story, which has been abysmal even after this story was published.
They could have taken auditions from 50 voice actors, come across this one, thought to themselves "Hey, this sounds just like the actress in 'Her', great, let's use them" and that would be fine. Laurence Fishburne does not own his "welcome to the desert of the real" intonation; other people have it too, and they can be hired to read in it.
Again: the Post has this voice actor reading in the normal voice. This wasn't an impersonator.
9 replies →
It was not claimed that they cloned ScarJo's voice. They hired a soundalike when they couldn't get the person they wanted. Use or lack of use of AI is irrelevant. As I said before, both Bette Midler and Tom Waits won similar cases.
Since they withdrew the voice this will end, but if OpenAI hadn't backed off and ScarJo sued, there would be discovery, and we'd find out what her instructions were. If those instructions were "try to sound like the AI in the film Her", that would be enough for ScarJo to win.
I know that the Post article claims otherwise. I'm skeptical.
> It was not claimed that they cloned ScarJo’s voice.
There were some claims by some people when the issue first arose that they had specifically done a deepfake clone of SJ’s voice; probably because of the combination of apparent trading on the similarity and the nature of OpenAI’s business. That’s not the case as far as the mechanism by which the voice was produced.
1 reply →
> They hired a soundalike when they couldn't get the person they wanted.
Your opinion may vary, but they don't sound alike to me: https://news.ycombinator.com/item?id=40435695
Your immediate acceptance that a timeline that represents the best spin of a deep-pocketed company in full crisis PR mode proves the story "false", full stop, no caveats is... I wouldn't say mind-bending, but quite credulous at a minimum. The timeline they present could be accurate but the full picture could still be quite damning. As Casey Newton wrote today [1]:
> Of course, this explanation only goes so far. We don’t know whether anyone involved in choosing Sky’s voice noted the similarity to Johansson’s, for example. And given how close the two voices sound to most ears, it might have seemed strange for the company to offer both the Sky voice and the Johansson voice, should the latter actor have chosen to participate in the project. [...] And I still don’t understand why Altman reportedly reached out to Johansson just two days before the demonstration to ask her to reconsider.
They absolutely have not earned the benefit of the doubt. Just look at their reaction to the NDA / equity clawback fiasco [2], and their focus on lifelong non-disparagement clauses. There's a lot of smoke there...
[1] https://www.platformer.news/openai-scarlett-johansson-chatgp...
[2] https://www.vox.com/future-perfect/351132/openai-vested-equi...
>They hired the actor that did the voice months before they contacted SJ. The reaction on this site to the news that this story was false is kind of mindbending.
People lose their rational mind when it comes to people they hate (or the opposite I suppose). I don't care for Sam Altman, or OpenAI one way or another, so it was quite amusing to watch the absolute outrage the story generated, with people so certain about their views.
I don't understand the point you are trying to make. The essential question is whether they were trying to imitate (using a voice actor or otherwise) Scarlett Johansson's voice without her permission. Nothing in the article refutes that they were; whether they sought the permission before or after they started doing the imitation is irrelevant. Others have pointed to previous case law that shows that this form of imitation is illegal.
Moreover I can't see any reasonable person concluding that they were not trying to imitate her voice given that:
1. It sounds similar to her (It's unbelievable that anyone would argue that they aren't similar, moreso given #2).
2. Her voice is famous for the context in which synthetic voice is used
3. They contacted her at some point to get her permission to use her voice
4. The CEO referenced the movie which Johansson's voice is famous for (and again depicts the same context the synthetic voice is being used) shortly before they released the synthetic voice.
Except the story isn't false? They wanted her voice, they got her voice*, they did marketing around her voice, but it's not her voice, she didn't want to give them her voice.
Notice how the only asterisk there is "it's technically not her voice, it's just someone who they picked because she sounded just like her"
>> They hired the actor that did the voice months before they contacted SJ.
Are you saying that story is false?
Yeah, but then again, I totally expected this opening the comment threads. Same happened with RMS debacle, same happened with similar events earlier, same happened on many a Musk stories. It seems that a neat narrative with clear person/object to hate, once established, is extremely resilient to facts that disprove it.
Right. Even if you think OpenAI isn’t a good place, this is an investigation by an established newspaper that refuted some of the more serious accusations (that OpenAI got a Johannson impersonator - they didn’t, that they modified the voice to sound like Johansson - evidence suggests this didn’t happen). When the reaction is “I don’t care that an investigation refuted some of the accusations”, it demonstrates someone isn’t openly approaching things in good faith.
Likewise, if someone’s attitude is - “OK, maybe there’s no paper trail, but I’m sure this is what the people were thinking”, then you’ve made an accusation that simply can’t be refuted, no matter how much evidence gets presented.
8 replies →
What facts disprove OpenAI making a voice that sounds like SJ such that the movie Her is referenced by Altman, and why is that actress upset?
4 replies →
I also see this dynamic on these same kinds of threads, but what I see is that one side is very sure that the facts disprove something, and the other side is very sure they don't. I've been on both sides of this, on different questions. I don't think there is anything weird about this, it's just a dispute over what a given fact pattern demonstrates. It's totally normal for people to disagree about that. It's why we put a fairly large number of people on a jury... People just see different things differently.
5 replies →
I'm not sure what RMS has to do with Altman. I'm also not sure why you think people just want to hate on Musk when it took a decade of his blatant lies for most people to catch on to the fact that he's a conman (remember, everyone loved him and Tesla for the first 5 or 10 years of lies). But the comparison between Musk and Altman is pretty apt, good job there.
4 replies →
It’s human nature: people see others achieve what they cannot, and try to pull them down. You see this wrt Musk on this site a lot, too.
It has nothing to do with this. There are many successful people and businesses that I admire, and a number of notable examples of those I do not. The two your comment mentions are simply part of that latter group. I think for good reason. (But of course I would think that...)
13 replies →
That was what Elizabeth Holmes claimed as well, however, we know that some people who try to achieve greatness are grifters. A pithy saying doesn’t change that reality.
3 replies →
Tbf here Altman really screwed this over with that tweet and very sudden contacting. There probably wouldn't be much of a case otherwise.
If I had to guess the best faith order of events (more than what OpenAi deserves):
- someone liked Her (clearly)
- they got a voice that sounded like Her, subconsciously (this is fine)
- someone high up hears it and thinks "wow this sounds like SJ!" (again, fine)
- they think "hey, we have money. Why not get THE SJ?!"
- they contact SJ, she refuses and they realize money's isn't enough (still fine. But this is definitely some schadenfreude here)
- marketing starts semi-indepenently, and they make references to Her, becsuse famous AI voice (here's where the cracks start to form. Sadly the marketer may not have even realized what talks went on).
- someone at OpenAi makes one last hail Mary before the release and contacts SJ again (this is where the trouble starts. MAYBE they didn't know about SJ refusing, but someone in the pipeline should have)
- Altman, who definitely should have been aware of these contacts, makes that tweet. Maybe they forgot, maybe they didn't realize the implications. But the lawyer's room is now on fire
So yeah, hanlon's razor. Thus could he a good faith mistake, but OpenAi's done a good job before this PR disaster ruining their goodwill. Again, sweet Schadenfreude even if we are assuming none of this was intentional.
Just how many "Good faith mistakes" is a company / CEO permitted to make before a person stops believing the good faith part?
2 replies →
If they gor the voice anyway why did they contact her?
The population of this site reacts to all stories like this. It’s only Gell-Mann Amnesia that causes your mind to bend.
thats even extra flattery for her
Legally, the issue isn’t what they were thinking when they hired the actor, it’s what the intent and effect was when they went to market. (Even if there was documentary evidence that they actively sought out an actor for resemblance to SJ’s voice from day one, the only reason that would be relevant is because it would also support that that was there intent with the product when it was actually released, not because it is independently relevant on its own.)
Whether or not they had any interest in SJ’s voice when they hired the other actor, they clearly developed such an interest before they went to market, and there is at least an evidence-based argument that could be made in court that they did, in fact, commercially leverage similarity.
No, that's not the legal standard here.
It is a curious reaction, but it starts to make sense if some of these posters are running ops for intelligence agencies. Balaji Srinivasan noted that as the US started pulling out of foreign wars, the intelligence apparatus would be turned inward domestically.
Some of it can also be attributed to ideological reasons, the d/acc crowd for example. Please note I am not attacking any individual poster, but speculating on the reasons why someone might refuse to acknowledge the truth, even when presented evidence to the contrary.
> Are we supposed to believe that the company lawyer or head of HR did this without consulting (or more likely being instructed by) the co-founder and CEO?!
Yes this is pretty typical. The CEO doesn’t make all decisions. They hire people to make decisions. A company’s head of legal could definitely make decisions about what standard language to use in documents on their own.
It could have simply been the other way around: they auditioned some unknown voice actors, then someone noted that one of them sounded like Scarlett Johansson. They optimistically contacted SJ, assuming she would agree, but then had to back off.