Scarlett Johansson Exposes OpenAI’s Rotten Business

OpenAI, the makers of ChatGPT, is a shitty company whose business model relies on copying other people’s content and calling it their own work, even going as far as to try to steal Scarlett Johansson’s voice. The world-famous actress said that she turned down OpenAI’s offer to pay her to voice their new chatbot called Sky, and then subsequently had her lawyers contact OpenAI and its CEO Sam Altman when Sky’s voice sounded just like hers.

“When I heard the released demo, I was shocked, angered and in disbelief that Mr. Altman would pursue a voice that sounded so eerily similar to mine that my closest friends and news outlets could not tell the difference. Mr. Altman even insinuated that the similarity was intentional, tweeting a single word ‘her’ — a reference to the film in which I voiced a chat system, Samantha, who forms an intimate relationship with a human.”

OpenAI claimed that they were not trying to make it sound like Scarlett Johansson, but the fact that they pulled it down betrays any kind of conviction they may have in that belief. This is just another day in the life of a garbage business whose CEO is full of shit, and their entire model relies on being able to farm content for free. They have already been sued to the tune of $3 billion, and this latest development with Scarlett Johansson indicates that they are eager to invite more litigation upon themselves.

It may sound strange to say a company that secured a recent investment valuing them at $80 billion is a crap business, but just because desperate venture capitalists who don’t want to miss out on the AI hype cycle are paying an expensive multiple does not mean that is what this company is actually worth.

Finding net income figures for OpenAI is a bit difficult, but revenue is not hard to locate at all (hmmmm). Two sources told the Financial Times earlier this year that the company surpassed $2 billion in revenues and that figure is cited in practically every story about the business. They are a private enterprise, so they are not required to make public filings of their financials, but The Information reported that net income was negative $540 million in 2022 according to three sources.

While the $2 billion in revenue for 2023 is one of the most readily available figures on the internet, 2023 net income is impossible to find, which definitely got the attention of my bullshit detector.

It’s nothing new for a young company to lose money, especially one who transitioned from non-profit status in the past year after a huge fight led to a shakeup on their board of directors. The standard cash flow pattern for any business is negative flows early on, leading to positive flows later once a consistent revenue stream has been established. OpenAI has a revenue stream, although whether it is consistent is impossible to tell until we get out of this AI hype cycle where everyone and their mother is trying to pretend that it’s a central aspect of their business.

The jump in revenues from 2022 to 2023 only indicates that customers and businesses are willing to try out OpenAI’s services, but demonstrating that those revenues are sustainable will take many more years to prove.

It is clear as day to anyone not brain poisoned by this hype cycle that OpenAI is a marketing scheme as much as it is an actual business at this present moment. They very likely scraped the entire internet to train their AI, and they know that admitting to it opens them up to all sorts of potential liabilities, as demonstrated by this painfully awkward non-response from their Chief Operating Officer to the question of whether they used YouTube to train Sora, their video and image-based AI.

OpenAI still can’t answer a simple yes or no question of whether YouTube videos were used to train Sora.

What you get instead is a total dodge and, “We’re looking at this problem, it’s really hard.”

They’re going to run out of executives to embarrass at this rate. pic.twitter.com/vOZYoU5iK7

— Reid Southen (@Rahll) May 10, 2024

Pro-tip: when an executive gives a meandering non-answer like this to an extremely straight-forward yes or no question, the answer is yes, they very obviously did use YouTube data to train Sora and they know that if they admitted that fact it would ruin a significant chunk of their business by actually having to pay for content intrinsic to their operations. The caginess of OpenAI’s executives whenever the subject comes up of how much content they scrape off the web just goes to show how they are aware that this is a legal grey area they are hoping will work out for them over time.

If it doesn’t, well that’s a problem for OpenAI’s later investors like Microsoft and its debtors to deal with. Everyone who was early already got paid. Welcome to venture capital.

OpenAI simply could not exist in its current form without taking people’s content for free, and they decided to highlight this for all to see by copying Scarlett Johansson’s voice after she rejected their requests to use it. This provides a window into the mindset of Silicon Valley leeches like Sam Altman, as these overgrown manchildren have their heads shoved so far up their own asses they’ve lost all contact with objective reality, as demonstrated by the fact that they believed they could just steal from one of the most famous people alive and get away with it.

Altman said that OpenAI needs the equivalent of one-third of annual United States GDP to invest in computer chips that can make their product work the way they want it to. That he can just throw a $7 trillion figure like that out there as if it is a normal request is a good measurement of how far detached he is from reality. Altman makes serial liar Elon Musk look grounded by comparison.

That said, I can understand how you would be worried about the future of your company and the need to raise a lot more cash when your latest innovation is being polluted by spam and porn websites while plenty of AI experts warn about the future degradation of AI as it trains itself on the rising tide of AI slop overtaking the web.

if you ever need a little confidence boost just remember you can always play rock paper scissors with chatgpt pic.twitter.com/XpUF4JqCTJ

— gabe (@allgarbled) May 21, 2024

A $7 trillion demand like this indicates that deep down, Altman knows that his business is not actually worth $80 billion. If he needs 87.5 times what the company is supposedly worth to make the product usable, then his product simply does not work according to his own logic. I’ve already detailed how this stuff is the furthest thing from “intelligent” that you can imagine, and I can only assume that a part of Altman believes that too given this batshit insane request for an investment that is twice the United Kingdom’s annual GDP.

OpenAI has already run afoul of European data protection laws, and nervous answers like the one given in the video above prove how fearful company executives are of being held accountable over their business model of taking other people’s content for free and calling it their own work. That Altman and OpenAI were willing to steal Scarlett Johansson’s voice even after she rejected their offer just shows how far above the law they believe themselves to be.

Perhaps YouTube should look into how much their product was involved in training Sora, at the very least it would be fun to watch Sam Altman squirm trying to assert that their core business is not dependent on stealing other people’s work when all available evidence indicates otherwise.