You can read the actual courthouse complaint: https://www.courthousenews.com/wp-content/uploads/2024/02/musk-v-altman-openai-complaint-sf.pdf Elon Musk founded OpenAI in 2015 with Sam Altman and...
Elon Musk founded OpenAI in 2015 with Sam Altman and Dario Amodei to develop artificial general intelligence (AGI) that would benefit humanity, not for-profit companies.
OpenAI was established as a non-profit with the goal of open-sourcing its technology when possible.
In 2020, OpenAI licensed its GPT-3 language model exclusively to Microsoft, going against its mission.
By 2023, Microsoft had influence over OpenAI's board and research directions through its funding of OpenAI's for-profit arm.
OpenAI's latest models like GPT-4 and Q* may constitute AGI but are being developed for Microsoft's profits rather than public benefit.
The complaint alleges OpenAI breached its founding agreement and fiduciary duties by prioritizing profits over its nonprofit mission.
It seeks damages, injunctions against improper use of funds, and a determination that GPT-4/Q* are outside Microsoft's license.
I think Musk is doing two interesting things.
First he is challenging OpenAIs unusual move to have a for profit subdivision own the tech that the nonprofit originally developed.
Second Musk is saying because GPT 4 is AGI, Microsoft has no rights to OpenAIs models.
I support both these positions, even though I don't support Musk.
I don't believe he will be successful proving the second point, but I think the first is likely to be interesting, Musk is the best person to file this claim as he personally was harmed by this move (so I think he has legal standing), he invested in a non profit and now Microsoft has control because all the employees are paid in for profit stock options or shares.
OpenAI went from transparency to secrecy, shutting the public out of its most advanced technology.
The changes have undermined OpenAI's original promises and raised concerns about its commitment to safety.
If Musk can claim GPT is AGI he will be successful in negating the current contract that entitles Microsoft to access GPT 4 technology. But I don't think he will be successful in proving GPT4 is...
I don't believe he will be successful proving the second point
If Musk can claim GPT is AGI he will be successful in negating the current contract that entitles Microsoft to access GPT 4 technology. But I don't think he will be successful in proving GPT4 is AGI. The courts will struggle to define AGI. They struggle with the definition of pornography. When the courts struggle, they typically say it's a problem for the legislators.
I'd be curious if anyone has further insight into this. On one hand, Musk may see OpenAi as a competition, either to his profit or his celebrity. On the other, he's insane and unpredictable so...
I'd be curious if anyone has further insight into this.
On one hand, Musk may see OpenAi as a competition, either to his profit or his celebrity.
On the other, he's insane and unpredictable so perhaps he thinks he believes what he's saying.
I think it's a mix of envy (that OpenAI gets a lot of media attention, is potentially seen as 'more innovative' than him) and anger (that the money he initially contributed has kind of wound up in...
I think it's a mix of envy (that OpenAI gets a lot of media attention, is potentially seen as 'more innovative' than him) and anger (that the money he initially contributed has kind of wound up in an essentially for-profit company which wasn't the original intention). I'm just grabbing some popcorn because I think this will be fun to watch.
Yea, as much as I am irritated by Musk, OpenAI has really taken a bad turn from the original promise. Mozilla seems to be doing good work in that vein, maybe he should throw a few billions that way.
Yea, as much as I am irritated by Musk, OpenAI has really taken a bad turn from the original promise.
Mozilla seems to be doing good work in that vein, maybe he should throw a few billions that way.
True, although I think Mozilla is spending too much money on political advocacy (which is kind of better done by EFF) and not enough money on making Firefox a faster/better/less buggy browser.
True, although I think Mozilla is spending too much money on political advocacy (which is kind of better done by EFF) and not enough money on making Firefox a faster/better/less buggy browser.
Somebody has talked about it before, but it has something to do with the way they are structured, they can't just pour all the money in the browser. TBH, the only times I've noticed Firefox being...
Somebody has talked about it before, but it has something to do with the way they are structured, they can't just pour all the money in the browser.
TBH, the only times I've noticed Firefox being slow or buggy is when I get a popup that says 'This page is only supported on Chrome.' Gives me IE PTSD. Generally also a heavily bloated page that's loading 20MB of analytics and ads instead of showing me my bank balance.
The question I have is why the foundation can’t fund open source development (of Firefox, in this case) as part of its “charitable mission.” It seems like there are other charities that do pay...
The question I have is why the foundation can’t fund open source development (of Firefox, in this case) as part of its “charitable mission.” It seems like there are other charities that do pay open source developers?
Maybe this is loopholey enough that it's illegal, but it seems like the nonprofit could develop everything and the Corporation could be the one that releases the product Firefox. So everything on...
Maybe this is loopholey enough that it's illegal, but it seems like the nonprofit could develop everything and the Corporation could be the one that releases the product Firefox. So everything on the backend is done by the Foundation while the Corporation packages it together as Firefox. It's all open source so anyone can make their own version. I don't see why that couldn't include the Corporation.
I just came across people discussing this on hackernews and did remember your awesome comments here. I hope you don't mind that I copied bits and pieces over in my reply there:...
I just came across people discussing this on hackernews and did remember your awesome comments here. I hope you don't mind that I copied bits and pieces over in my reply there: https://news.ycombinator.com/item?id=39589402
The most interesting thing here to me is that Musk is claiming that GPT-4 is in fact a form of AGI, which is perhaps somewhat legally supported by Microsoft releasing a paper claiming that it...
The most interesting thing here to me is that Musk is claiming that GPT-4 is in fact a form of AGI, which is perhaps somewhat legally supported by Microsoft releasing a paper claiming that it contains the "sparks" of AGI. The reason this is important is,
Microsoft continues to assert rights to GPT-4, which it claims has not reached the level of AGI, which would block its licensing privileges.
I'm curious what other consequences would entail from a ruling that GPT-4 is a form of AGI. To be clear, I don't at all believe GPT-4 to be AGI.
I think the "LLMs are AGI" take is nonsense, and that Microsoft paper doesn't consider the fact that a lot of the data they are evaluating the model on is actually in the training dataset! There...
I think the "LLMs are AGI" take is nonsense, and that Microsoft paper doesn't consider the fact that a lot of the data they are evaluating the model on is actually in the training dataset!
I don’t disagree, but the courts have not always shown themselves to be technologically competent. Once in a while we get a notable judge like the one who was involved in the Java case, but a...
I don’t disagree, but the courts have not always shown themselves to be technologically competent. Once in a while we get a notable judge like the one who was involved in the Java case, but a large number of people are caught up in the hype.
My thinking here though, is that in terms of a court case, Microsoft is making what seems to be an admission that it is AGI (or sparks of), which at some level contradicts their other claim.
This article seems to lack details on the most important part of a meritorious lawsuit: actual damages. So if anyone else is wondering what damages Musk alleges to have suffered, you can find them...
This article seems to lack details on the most important part of a meritorious lawsuit: actual damages. So if anyone else is wondering what damages Musk alleges to have suffered, you can find them around paragraphs 126, 132, 135, and 143 of the complaint.
Rather than quote like fifteen paragraphs, I'll just summarize the idea here. Basically, Musk claims five causes of action against OpenAI:
Breach of contract,
Promissory estoppel,
Breach of fiduciary duty,
Unfair business practices,
Accounting
But all of these cause of actions amount to essentially the same thing: Musk contributed money and advice to OpenAI under the assumption that OpenAI's research would be used in accordance with the "Founding Agreement", which the complaint describes as a pledge that
OpenAI, Inc. (a) would be a non-profit developing AGI for the benefit of humanity, not for a for-profit company seeking to maximize shareholder profits; and (b) would be open-source, balancing only countervailing safety considerations, and would not keep its technology closed and secret for proprietary commercial reasons.
As for the amount of damages suffered, the complaint has this to say:
As a direct and proximate result of Defendants’ breaches, Plaintiff has suffered damages in an amount that is presently unknown, but that substantially exceeds this Court’s jurisdictional minimum of $35,000, and, if necessary, will be proven at trial.
Uh, yes, that would be necessary?
To be honest, I don't think the lawsuit makes a very compelling case that Musk actually suffered damages? For one, OpenAI is not a for-profit company but rather a capped-profit company as Matt Levine humorously pointed out in his newsletter. You could argue that this is a distinction without a difference, but the counterargument is that companies need money to attract world-class talent, and non-profits do not traditionally have world-class-talent money.
But the deeper problem is that a pledge to "develop[e] AGI for the benefit of humanity" is not, like, a legal term of art. It's more like a mantra. Musk can disagree with OpenAI's business strategy, but typically, so long as a decision is reasonable, a shareholder cannot sue a company just because they would have done something differently. And Musk isn't even a shareholder. I don't see how the "breach of fiduciary duty" claim has any legs because I don't see how OpenAI has a fiduciary duty to Musk.
Also, for those that missed it, the prayer for relief makes a pretty bold ask:
B. For a judicial determination that GPT-4 constitutes Artificial General Intelligence and is thereby outside the scope of OpenAI’s license to Microsoft;
Future headline: Supreme Court rules that Bard, ChatGPT constitute artificial general intelligence, Clippy does not.
I’m wondering how much Musk donated to OpenAI and whether they could settle this by paying him back? How much in damages is he asking and how is that justified?
I’m wondering how much Musk donated to OpenAI and whether they could settle this by paying him back? How much in damages is he asking and how is that justified?
I believe the answer is not public – Musk has previously stated US$100mn and more recently US$50mn. Elon Musk has a very niche and contrarian opinion when it come to legal disputes which usually...
I’m wondering how much Musk donated to OpenAI
I believe the answer is not public – Musk has previously stated US$100mn and more recently US$50mn.
whether they could settle this by paying him back?
Elon Musk has a very niche and contrarian opinion when it come to legal disputes which usually works out as "I have more money than you so I will keep fighting and do things which don't make financial sense for me because I find them fun until I crush you" so I kind of (based on past data) think this isn't super likely.
Oh he's totally doing it because he's jealous and they're a competitor to his new LLM (marketed essentially on the basis that it will spit out very right wing views).
Oh he's totally doing it because he's jealous and they're a competitor to his new LLM (marketed essentially on the basis that it will spit out very right wing views).
You can read the actual courthouse complaint:
https://www.courthousenews.com/wp-content/uploads/2024/02/musk-v-altman-openai-complaint-sf.pdf
I think Musk is doing two interesting things.
First he is challenging OpenAIs unusual move to have a for profit subdivision own the tech that the nonprofit originally developed.
Second Musk is saying because GPT 4 is AGI, Microsoft has no rights to OpenAIs models.
I support both these positions, even though I don't support Musk.
I don't believe he will be successful proving the second point, but I think the first is likely to be interesting, Musk is the best person to file this claim as he personally was harmed by this move (so I think he has legal standing), he invested in a non profit and now Microsoft has control because all the employees are paid in for profit stock options or shares.
OpenAI went from transparency to secrecy, shutting the public out of its most advanced technology.
The changes have undermined OpenAI's original promises and raised concerns about its commitment to safety.
How do you support claiming that GPT4 is an AGI?
If Musk can claim GPT is AGI he will be successful in negating the current contract that entitles Microsoft to access GPT 4 technology. But I don't think he will be successful in proving GPT4 is AGI. The courts will struggle to define AGI. They struggle with the definition of pornography. When the courts struggle, they typically say it's a problem for the legislators.
I'd be curious if anyone has further insight into this.
On one hand, Musk may see OpenAi as a competition, either to his profit or his celebrity.
On the other, he's insane and unpredictable so perhaps he thinks he believes what he's saying.
I think it's a mix of envy (that OpenAI gets a lot of media attention, is potentially seen as 'more innovative' than him) and anger (that the money he initially contributed has kind of wound up in an essentially for-profit company which wasn't the original intention). I'm just grabbing some popcorn because I think this will be fun to watch.
Yea, as much as I am irritated by Musk, OpenAI has really taken a bad turn from the original promise.
Mozilla seems to be doing good work in that vein, maybe he should throw a few billions that way.
True, although I think Mozilla is spending too much money on political advocacy (which is kind of better done by EFF) and not enough money on making Firefox a faster/better/less buggy browser.
Somebody has talked about it before, but it has something to do with the way they are structured, they can't just pour all the money in the browser.
TBH, the only times I've noticed Firefox being slow or buggy is when I get a popup that says 'This page is only supported on Chrome.' Gives me IE PTSD. Generally also a heavily bloated page that's loading 20MB of analytics and ads instead of showing me my bank balance.
I mean the corporation has $400mn in annual revenue or something and I think they'd find it tough to top that in donations.
Is there an article somewhere that explains this? It seems like most people don’t know it.
The question I have is why the foundation can’t fund open source development (of Firefox, in this case) as part of its “charitable mission.” It seems like there are other charities that do pay open source developers?
Maybe this is loopholey enough that it's illegal, but it seems like the nonprofit could develop everything and the Corporation could be the one that releases the product Firefox. So everything on the backend is done by the Foundation while the Corporation packages it together as Firefox. It's all open source so anyone can make their own version. I don't see why that couldn't include the Corporation.
I just came across people discussing this on hackernews and did remember your awesome comments here. I hope you don't mind that I copied bits and pieces over in my reply there: https://news.ycombinator.com/item?id=39589402
The most interesting thing here to me is that Musk is claiming that GPT-4 is in fact a form of AGI, which is perhaps somewhat legally supported by Microsoft releasing a paper claiming that it contains the "sparks" of AGI. The reason this is important is,
I'm curious what other consequences would entail from a ruling that GPT-4 is a form of AGI. To be clear, I don't at all believe GPT-4 to be AGI.
Additional coverage:
Courthouse News - Elon Musk sues OpenAI over AI threat
Reuters - Elon Musk sues OpenAI for abandoning original mission for profit
The Guardian - Elon Musk sues OpenAI accusing it of putting profit before humanity
I think the "LLMs are AGI" take is nonsense, and that Microsoft paper doesn't consider the fact that a lot of the data they are evaluating the model on is actually in the training dataset!
There are some great papers out there explaining the problems with LLMs, for example "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?". There's also a Twitter thread from one of the authors of that paper on the Microsoft one.
I don’t disagree, but the courts have not always shown themselves to be technologically competent. Once in a while we get a notable judge like the one who was involved in the Java case, but a large number of people are caught up in the hype.
My thinking here though, is that in terms of a court case, Microsoft is making what seems to be an admission that it is AGI (or sparks of), which at some level contradicts their other claim.
This article seems to lack details on the most important part of a meritorious lawsuit: actual damages. So if anyone else is wondering what damages Musk alleges to have suffered, you can find them around paragraphs 126, 132, 135, and 143 of the complaint.
Rather than quote like fifteen paragraphs, I'll just summarize the idea here. Basically, Musk claims five causes of action against OpenAI:
But all of these cause of actions amount to essentially the same thing: Musk contributed money and advice to OpenAI under the assumption that OpenAI's research would be used in accordance with the "Founding Agreement", which the complaint describes as a pledge that
As for the amount of damages suffered, the complaint has this to say:
Uh, yes, that would be necessary?
To be honest, I don't think the lawsuit makes a very compelling case that Musk actually suffered damages? For one, OpenAI is not a for-profit company but rather a capped-profit company as Matt Levine humorously pointed out in his newsletter. You could argue that this is a distinction without a difference, but the counterargument is that companies need money to attract world-class talent, and non-profits do not traditionally have world-class-talent money.
But the deeper problem is that a pledge to "develop[e] AGI for the benefit of humanity" is not, like, a legal term of art. It's more like a mantra. Musk can disagree with OpenAI's business strategy, but typically, so long as a decision is reasonable, a shareholder cannot sue a company just because they would have done something differently. And Musk isn't even a shareholder. I don't see how the "breach of fiduciary duty" claim has any legs because I don't see how OpenAI has a fiduciary duty to Musk.
Also, for those that missed it, the prayer for relief makes a pretty bold ask:
Future headline: Supreme Court rules that Bard, ChatGPT constitute artificial general intelligence, Clippy does not.
Elon Musk’s legal case against OpenAI is hilariously bad
I’m wondering how much Musk donated to OpenAI and whether they could settle this by paying him back? How much in damages is he asking and how is that justified?
I believe the answer is not public – Musk has previously stated US$100mn and more recently US$50mn.
Elon Musk has a very niche and contrarian opinion when it come to legal disputes which usually works out as "I have more money than you so I will keep fighting and do things which don't make financial sense for me because I find them fun until I crush you" so I kind of (based on past data) think this isn't super likely.
I believe the complaint is public at this point but haven’t had time to read it. I’ve seen 44 million thrown around
One of those cases where I agree with the message (headline at least), but the messenger and his possible motives rub me the wrong way
Oh he's totally doing it because he's jealous and they're a competitor to his new LLM (marketed essentially on the basis that it will spit out very right wing views).