Also, what if you can't sell? Selling is at their discretion. They can prevent you from selling some of your so-called "equity" to keep you on their leash as long as they want.
Well, that would obviously depend on the terms of the contract, but I would be astonished if the people who wrote it didn't consider that possibility. It's pretty trivial to calculate the monetary value of equity, and if they feel entitled to that equity, they surely feel entitled to its cash equivalent.
Probably people like Kokotajlo cared about the value of their equity but even more about their other principles, like speaking the truth publicly even if it meant their losing millions.
Wait that's a thing? Can you give more detail about this/what to look into to learn more?
>” The Silenced No More Act bans confidentiality provisions in settlement agreements relating to the disclosure of underlying factual information relating to any type of harassment, discrimination or retaliation at work”
I find it hard to understand that in a country that tends to take freedom of expression so seriously (and I say this unironically, American democracy may have flaws but that is definitely a strength) it can be legal to silence someone for the rest of their life.
Its quite common for companies to put tons of extremely restrictive terms in an NDA they can't actually legally enforce to scare off potential future ex-employees from creating a problem.
From the article:
“““
It turns out there’s a very clear reason for [why no one who had once worked at OpenAI was talking]. I have seen the extremely restrictive off-boarding agreement that contains nondisclosure and non-disparagement provisions former OpenAI employees are subject to. It forbids them, for the rest of their lives, from criticizing their former employer. Even acknowledging that the NDA exists is a violation of it.
If a departing employee declines to sign the document, or if they violate it, they can lose all vested equity they earned during their time at the company, which is likely worth millions of dollars. One former employee, Daniel Kokotajlo, who posted that he quit OpenAI “due to losing confidence that it would behave responsibly around the time of AGI,” has confirmed publicly that he had to surrender what would have likely turned out to be a huge sum of money in order to quit without signing the document.
”””
[0]: https://www.vox.com/future-perfect/2024/5/17/24158478/openai...
It takes a man of real principle to stand up against that and tell them to keep their money if they can't speak ill of a potentially toxic work environment.
Incidentally, that's what Grigory Perelman, the mathematician that rejected the Fields Medal and the $1M prize that came with it, did.
It wasn't a matter of an NDA either; it was a move to make his message heard (TL;DR: "publish or perish" rat race that the academia has become is antithetical to good science).
He was (and still is) widely misunderstood in that move, but I hope people would see it more clearly now.
The enshittification processes of academic and corporate structures are not entirely dissimilar, after all, as money is at the core of corrupting either.
This is the kind of thing a cult demands of its followers, or an authoritarian government demands of its citizens. I don't know why people would think it's okay for a business to demand this from its employees.
Perfect! So it's so incredibly overreaching that any judge in California would deem the entire NDA unenforceable..
Either that or, in your effort to overstate a point, you exaggerated in a way that undermines the point you were trying to make.
This should not be legal.
Idk. Folks much smarter than I seem worried so maybe I should be too but it just seems like such a long shot.
So yes, the insiders very likely know a thing or two that the rest of us don’t.
That's easy, we just need to make meatspace people stupider. Seems to be working great so far.
Honestly? I'm not too worried
We've seen how the google employee that was "seeing a conscience" (in what was basically GPT-2 lol) was a nothing burger
We've seen other people in "AI Safety" overplay their importance and hype their CV more than actually do any relevant work. (Usually also playing the diversity card)
So, no, AI safety is important but I see it attracting the least helpful and resourceful people to the area.
Given the model is probabilistic and does many things in parallel, its output can be understood as a mixture, e.g. 30% trash, 60% rehashed training material, 10% reasoning.
People probe model in different ways, they see different results, and they make different conclusions.
E.g. somebody who assumes AI should have impeccable logic will find "trash" content (e.g. incorrectly retrieved memory) and will declare that the whole AI thing is overhyped bullshit.
Other people might call model a "stochastic parrot" as they recognize it basically just interpolates between parts of the training material.
Finally, people who want to probe reasoning capabilities might find it among the trash. E.g. people found that LLMs can evaluate non-trivial Python code as long as it sends intermediate results to output: https://x.com/GrantSlatton/status/1600388425651453953
I interpret "feel the AGI" (Ilya Sutskever slogan, now repeated by Jan Leike) as a focus on these capabilities, rather than on mistakes it makes. E.g. if we go from 0.1% reasoning to 1% reasoning it's a 10x gain in capabilities, while to an outsider it might look like "it's 99% trash".
In any case, I'd rather trust intuition of people like Ilya Sutskever and Jan Leike. They aren't trying to sell something, and overhyping the tech is not in their interest.
Regarding "missing something really critical", it's obvious that human learning is much more efficient than NN learning. So there's some algorithm people are missing. But is it really required for AGI?
And regarding "It cannot reason" - I've seen LLMs doing rather complex stuff which is almost certainly not in the training set, what is it if not reasoning? It's hard to take "it cannot reason" seriously from people
Nobody defines what they’re trying to do as “useful AI” since that’s a much more weasily target, isn’t it?
The whole industry at this point is acting like the tobacco industry back when they first started getting in hot water. No doubt the prophecies about imminent AGI will one day look to our descendents exactly like filters on cigarettes. A weak attempt to prevent imminent regulation and reduced profitability as governments force an out of control industry to deal with the externalities involved in the creation of their products.
If it wasn't abundantly clear...I agree with you that AGI is infinitely far away. Its the damage that's going to be caused by sociopaths (Sam Altman at the top of the list) in attempting to justify the real things they want (money) in their march towards that impossible goal that concerns me.
I think it may time for something like this: https://www.openailetter.org/
https://nitter.poast.org/janleike/status/1791498174659715494
All due respect to Jan here, though. He's being (perhaps dangerously) honest, genuinely believes in AI safety, and is an actual research expert, unlike me.
https://openai.com/index/introducing-superalignment/
> Superintelligence will be the most impactful technology humanity has ever invented, and could help us solve many of the world’s most important problems. But the vast power of superintelligence could also be very dangerous, and could lead to the disempowerment of humanity or even human extinction.
> While superintelligence seems far off now, we believe it could arrive this decade.
> Managing these risks will require, among other things, new institutions for governance and solving the problem of superintelligence alignment:
> How do we ensure AI systems much smarter than humans follow human intent?
> Currently, we don't have a solution for steering or controlling a potentially superintelligent AI, and preventing it from going rogue. Our current techniques for aligning AI, such as reinforcement learning from human feedback, rely on humans’ ability to supervise AI. But humans won’t be able to reliably supervise AI systems much smarter than us, and so our current alignment techniques will not scale to superintelligence. We need new scientific and technical breakthroughs.
OpenAI made a large commitment to super-alignment in the not-so-distant past. I beleive mid-2023. Famously, it has always taken AI Safety™ very seriously.
Regardless of anyone's feelings on the need for a dedicated team for it, you can chalk to one up as another instance of OpenAI cough leadership cough speaking out of both sides of it's mouth as is convenient. The only true north star is fame, glory, and user count, dressed up as humble "research"
To really stress this: OpenAI's still-present cofounder shared yesterday on a podcast that they expect AGI in ~2 years and ASI (superpassing human intelligence) by end of the decade.
How can I be confident you aren't committing the fallacy of collecting a bunch of events and saying that is sufficient to serve as a cohesive explanation? No offense intended, but the comment above has many of the qualities of a classic rant.
If I'm wrong, perhaps you could elaborate? If I'm not wrong, maybe you could reconsider?
Don't forget that alignment research has existed longer than OpenAI. It would be a stretch to claim that the original AI safety researchers were using the pretexts you described -- I think it is fair to say they were involved because of genuine concern, not because it was a trendy or self-serving thing to do.
Some of those researchers and people they influenced ended up at OpenAI. So it would be a mistake or at least an oversimplification to claim that AI safety is some kind of pretext at OpenAI. Could it be a pretext for some people in the organization, to some degree? Sure, it could. But is it a significant effect? One that fits your complex narrative, above? I find that unlikely.
Making sense of an organization's intentions requires a lot of analysis and care, due to the combination of actors and varying influence.
There are simpler, more likely explanations, such as: AI safety wasn't a profit center, and over time other departments in OpenAI got more staff, more influence, and so on. This is a problem, for sure, but there is no "pearl clutching pretext" needed for this explanation.
Care to explain? Absurd how? An internal contradiction somehow? Unimportant for some reason? Impossible for some reason?
1)OpenAI wouldn't want the negative PR of pursuing legal action against someone top in their field; his peers would take note of it and be less willing to work for them.
2)The stuff he signed was almost certainly different from what rank and file signed, if only because he would have sufficient power to negotiate those contracts.
Delusional.
Large language models are not "smart". They do not have thought. They don't have intelligence despite the "AI" moniker, etc.
They vomit words based off very fancy statistics.
There is no path from that to "thought" and "intelligence."
Maybe the agreement is "we will accelerate vesting of your unvested equity if you sign this new agreement"? If that's the case then it doesn't sound nearly so coercive to me.
> They don't share the equivalent of a Cap Table with employees, so there's no way to tell what sort of ownership interest a PPU represents
It is known - it represents 0 ownership share. They do not want to sell any ownership because their deal with MS gives MS 49% ownership and they don't want MS to be able to buy up additional stake and control the company.
> And of course, it's unlikely OpenAI will ever turn a profit (which if they did would be capped anyway). So this is all just play money anyway.
Putting aside your unreasonable confidence that OAI will never be profitable, the PPUs are tender offered so they can be sold to institutional investors up to a very high limit, OAIs current tender offer round values them at ~$80b iirc
They're really lending employees equity, subject to the company's later feelings as to whether the employee should be allowed to keep or sell it.
They're not required to sign anything other than a general release of liability when they leave to preserve their rights. They don't have to sign a non-disparagement clause.
But they'd need a very good lawyer to be confident at that time.
In general, an agreement to agree is not an agreement. A requirement for a "general release" to be signed at some time in the future is iffy. And that's before labor law issues.
Someone with a copy of that contract should run it through OpenAI's contract analyzer.
What a horrific medium of communication. Why anyone uses it is beyond me.
all this said, in bigger picture I can understand not divulging trade secrets but not being allowed to discuss company culture towards AI safety essentially tells me that all the Sama talk about the 'for the good of humanity' is total BS. at the end of day its about market share and bottom line.
Equity adds a wrinkle here, but I suspect if the effect of canceling equity is to cause a forfeiture of earned wages, then ultimately whatever contract is signed under that threat is void.
(Don’t have X) - is there a timeline? Can I curse out the company on my deathbed, or would their lawyers have the legal right to try and clawback the equity from the estate?
As for other companies that can pay: I can only assume that the cost to bribe skilled workers isn't worth the perceived risk and cost of lawsuits from the downfall (which they may or may not be able to settle). Generative AI is still very young and under a lot of scrutiny on all fronts, so the risk of a whistle blower at this stage may shape the entire future of the industry at large.
https://www.vox.com/future-perfect/2024/5/17/24158478/openai...
This thread is full of comments making statements around this looking like some level of criminal enterprise (ranging from “no way that document holds up” to “everyone knows Sam is a crook”).
The level of stuff ranging from vitriol to overwhelming if maybe circumstantial (but conclusive that my personal satisfaction) evidence of direct reprisal has just been surreal, but it’s surreal in a different way to see people talking about this like it was never even controversial to be skeptical/critical/hostile to thing thing.
I’ve been saying that this looks like the next Enron, minimum, for easily five years, arguably double that.
Is this the last straw where I stop getting messed around over this?
I know better than to expect a ticker tape parade for having both called this and having the guts to stand up to these folks, but I do hold out a little hope for even a grudging acknowledgment.
What made you think it was the next Enron five years ago?
I appreciate you having the guts to stand up to them.
Employees and employer enter into an agreement: Work here for X term and you get Y options with Z terms attached. OK.
But, then later pulling Darth Vader… “Now that the deal is completing, I am changing the deal. Consent and it’s bad for you this way. Don’t consent and it’s bad that way. Either way, you held up your end of our agreement and I’m not.”
(I should reiterate that I actually wrote "serious, possibly criminal")
So maybe it's related to that.
They will have many successes in the short run, but, their long run future suddenly looks a little murky.
>in regards to recent stuff about how openai handles equity:
>we have never clawed back anyone's vested equity, nor will we do that if people do not sign a separation agreement (or don't agree to a non-disparagement agreement). vested equity is vested equity, full stop.
>there was a provision about potential equity cancellation in our previous exit docs; although we never clawed anything back, it should never have been something we had in any documents or communication. this is on me and one of the few times i've been genuinely embarrassed running openai; i did not know this was happening and i should have.
>the team was already in the process of fixing the standard exit paperwork over the past month or so. if any former employee who signed one of those old agreements is worried about it, they can contact me and we'll fix that too. very sorry about this. https://x.com/sama/status/1791936857594581428
I hope ex-employees sue and don’t contact him personally. The damage is done. Don’t be dumb folks.
However they could add this to new employee contracts.
Pushing unenforceable scare-copy to get employees to self-censor sounds on-brand.
Doesn’t mean that that’s legal, of course, but I’d doubt that the legality would hinge on a lack of consideration.
If it was "we'll give you shares/cash if you don't say anything bad about us", that's normal, kind of standard fare for exit agreements, it's why severance packages exist.
But if it is "we'll take away the shares that you already earned as part of your regular employment compensation unless you agree to not say anything bad about us", that's extortion.
Many year ago I signed a NDA/non-disparagement agreement as part of a severance package when I was fired from a startup for political reasons. I didn't want to sign it... but my family needed the money and I swallowed my pride. There was a lot of unethical stuff going on within the company in terms of fiducial responsibility to investors and BoD. The BoD eventually figured out what was going on and "cleaned house".
With OpenAI, I am concerned this is turning into huge power/money grab with little care for humanity... and "power tends to corrupt and absolute power corrupts absolutely".
The power grab happened a while ago (the shenanigans concerning the board) and is now complete. Care for humanity was just marketing or a cute thought at best.
Maybe humanity will survive life long enough that a company "caring about humanity" becomes possible, I'm not saying it's not worth trying or aspiring to such ideals, but everyone should be extremely surprised if any organization managed to resist such amounts of money to maintain any goal or ideal whatever...
Sound like a better solution?
Well... I know first hand that many well-informed, tech-literate people still think that all products from OpenAI are open-source. Lying works, even in that most egregious of fashion.
Unfortunately Orwellian propoganda works.
Diamond multi-million dollar hand-cuffs which OpenAI has bound lifetime secret service-level NDAs which are another unusual company setting after their so-called "non-profit" founding and their contradictory name.
Even an ex-employee saying 'ClosedAI' could see their PPUs evaporate in front of them to zero or they could never be allowed to sell them and have them taken away.
Companies can cancel your vested equity for any reason. Read your employment contract carefully. For example, most RSU grants have a 7 year expiration. Even for shares that are vested, regardless of whether you leave the company or not, if 7 years have elapsed since they were granted, they are now worthless.
Once vested, RSUs are the same as regular stock purchased through the market. The company cannot claw them back, nor do they "expire".
> same as regular stock purchased through the market
You cannot purchase stock of a private company on the open market.
> The company cannot claw them back
The company cannot "claw back" a vested RSU but they can cancel it.
> nor do they "expire".
Yes, they absolutely do expire. Read your employment contract and equity grant agreement carefully.
But none of this means the company can just cancel your RSUs unless you agreed to them being cancelled for specific reason in your equity agreement. I have worked at several big pre-IPO companies that had big exits. I made sure there were no clawback clauses in the equity contract before accepting the offers.
Right. In the case of OpenAI, their equity grant contracts likely have a non-disparagement clause that allows them to cancel vested shares. Whether or not you think that is a "valid reason" is largely independent of the legal framework governing RSU release.
Turns out they're right, they can put whatever they want in a contract. And again, they are correct that their wage slaves will 99.99% of the time sign whatever paper he pushes in front of them while saying "as a condition of your continued employment, [...]".
But also it turns out that just because you signed something doesn't mean that's it. My friends (all of us young twenty-something software engineers much more familiar with transaction isolation semantics than with contract law) consulted with an attorney.
The TLDR is that:
- nothing in contract law is in perpetuity
- there MUST be consideration for each side (where "consideration" means getting something. something real. like USD. "continued employment" is not consideration.)
- if nothing is perpetual, then how long can it last supposing both sides do get ongoing consideration from it? the answer is, the judge will figure it out.
- and when it comes to employers and employees, the employee had damn well better be getting a good deal out of it, especially if you are trying to prevent the employee (or ex-employee) from working.
A common pattern ended up emerging: our employer would put something perpetual in the contract, and offer no consideration. Our attorney would tell us this isn't even a valid contract and not to worry about it. Employer would offer an employee some nominal amount of USD in severance and put something in perpetuity into the contract. Our attorney tells us the judge would likely use "blue ink rule" to add in "for a period of one year", or, it would be prorated based on the amount of money they were given relative to their former salary.
(I don't work there anymore, naturally).
Isn't that the reason more competent lawyers put in the royal lives[1] clause? It specifies the contract is valid until 21 years after the death of the last currently-living royal descendant; I believe the youngest one is currently 1 year old, and they all have good healthcare, so it's almost certainly will be beyond the lifetime of any currently-employed persons.
Would any non-corrupt judge consider this is done in bad fait?
How is this difference if we use a great ancient sea turtles—or some other long-lived organism—instead of the current royal family baby? Like, I guess my point is anything that would likely outlive the employee basically?
Even lowest level fast food workers can choose a different employer. An engineer working at OpenAI certainly has a lot of opportunities to choose from. Even when I only had three years in the industry, mid at best, I asked to change the contract I was presented with because non-compete was too restrictive — and they did it. The caliber of talent that OpenAI is attracting (or hopes to attract) can certainly do this too.
Only thanks to a recent ruling by the FTC that non-competes are valid. in the most egregious uses, bartenders and servers were prohibited from finding another job in the same industry for two years.
Personally I'd say there needs to be a general restriction against including blatantly unenforceable terms in a contract document, especially unilateral "terms". The drafter is essentially pushing incorrect legal advice.
Hmmmn. Most of the humans where I work do things physically with their hands. I don't see what AI will achieve in their area.
Can AI paint the walls in my house, fix the boiler and swap out the rotten windows? If so I think a subscription to chat GPT is very reasonably priced!
But if your job is mostly sitting at a computer, I would be a bit worried.
Now it’s a money grab.
Sad because some amazing tech and people now getting corrupted into a toxic culture that didn’t have to be that way
Hey hey hey! Sam founded a 4th most popular social networking site in 2005 called Loopt. Don't you forget that! (After that he joined YC and founded nothing ever since)
It's quite natural, that a co-founder, being forced out of the company wouldn't be exactly willing to forfeit his equity. So, what, now he cannot… talk? That has some Mexican cartel vibes.
If OpenAI and ChatGPT is so far ahead for everyone else, and their product is so complex, it doesn't matter what a few disgruntled employees do or say, so the rule is not required.
What's the consideration for this contract?
Everyone including the board's own chosen replacements for Altman siding with Altman seems to me to not be compatible with his current leadership being the root cause of the current discontent… so I'm blaming Microsoft, who were the moustache-twirling villains when I was a teen.
Of course, thanks to the NDAs hiding information, I may just be wildly wrong.
The most dishonest leadership.
There's more info on how SpaceX uses a scheme like this[0] to force compliance, and seeing as Musk had a hand in creating both orgs, they're bound to be similar.
[0] https://techcrunch.com/2024/03/15/spacex-employee-stock-sale...
Consideration is almost meaningless as an obstacle here. They can give the other party a peppercorn, and that would be enough to count as consideration.
https://en.wikipedia.org/wiki/Peppercorn_(law)
There might be other legal challenges here, but 'consideration' is unlikely to be one of them. Unless OpenAI has idiots for lawyers.
Speculation: maybe the options they earn when they work there have some provision like this. In return for the NDA the options get extended.
I’m not saying it’s right or that I agree with it, however.
> After publication, an OpenAI spokesperson sent me this statement: “We have never canceled any current or former employee’s vested equity nor will we if people do not sign a release or nondisparagement agreement when they exit.”
- Updated May 17, 2024, 11:20pm EDT
I've noticed that both Sam Altman personally, and official statements from OpenAI sound like they've been written by Aes Sedai: Not a single untrue word while simultaneously thoroughly deceptive.[1]
Let's try translating some statements, as if we were listening to an evil person that can only make true statements:
"We have never canceled any current or former employee’s vested equity" => "But we can and will if we want to. We just haven't yet."
"...if people do not sign a release or nondisparagement agreement when they exit." => "But we're making everyone sign the agreement."
[1] I've wondered if they use a not-for-public-use version of GPT for this purpose. You know, a model that's not quite as aligned as the chat bots, with more "flexible" morals.
I know many people on this site will not like what I am about to write as Sam is worshiped but let's face it: The head of this company is a master scammer who will do everything under the sun and the moon to earn a buck, including but notwithstanding to destroying himself along with his entire fortune if necessary in his quest of making sure other people don't get a dime;
So far he has done it all it: attempt to regulatory capture, hostile take over as the CEO, thrown out all other top engineers and partners and ensured the company remains closed despite its "open" name.
Now he is simply attempting to tie up all the loos ends and ensuring his employees remain loyal and are kept on a tight leash. It's a brilliant strategy, preventing any insider from blowing the whistle should OpenAI ever decides to do anything questionable, such as selling AI capabilities to hostile governments.
I simply hope that open source wins this battle so that we are not all completely reliant on OpenAI for the future, despite Sam's attempt.
This is the article that the author talks about on X.
If someone shares something that's a lie and defamatory, then they could still be sued of course.
The Ben Shapiro-Daily Wire vs. Candace Owens is another scenario where the truth and conversation would benefit all of society - OpenAI and DailyWire arguably being on topics of pinnacle importance; instead the discussions are suppressed.
It seems that standard practice would dictate that you sign an NDA before even signing the employment contract.
[1]https://www.lesswrong.com/posts/kovCotfpTFWFXaxwi/simeon_c-s...
There’s a very real/significant risk that AGI either literally destroys the human race, or makes life much shittier for most humans by making most of us obsolete. These risks are precisely why OpenAI was founded as a very open company with a charter that would firmly put the needs of humanity over their own pocketbooks, highly focused on the alignment problem. Instead they’ve closed up, become your standard company looking to make themselves ultra wealthy, and they seem like an extra vicious, “win at any cost” one at that. This plus their AI alignment people leaving in droves (and being muzzled on the way out) should be scary to pretty much everyone.
I'm not sure this is true. If all the things people are doing are done so much more cheaply they're almost free, that would be good for us, as we're also the buyers as well as the workers.
However, I also doubt the premise.
If this were true, intelligent people would have taken over society by now. Those in power will never relinquish it to a computer just as they refuse to relinquish it to more competent people. For the vast majority of people, AI not only doesn't pose a risk but will only help reveal the incompetence of the ruling class.
He clearly states why he left. He believes that OpenAI leadership is prioritizing shiny product releases over safety and that this is a mistake.
Even with the best intentions , it’s easy for a strong CEO like Altman to loose sight of more subtly important things like safety and optimize for growth and winning, eventually at all cost. Winning is a super-addictive feedback loop.
They can totally deal with appearing petty and thin-skinned.
So yes, they're that fragile.
It’s yet another sign that the AI bubble will soon burst. The laughable release of “GPT-4o” was just a small red flag.
Got to keep the soldiers in check while the bean counters prep the books for an IPO and eventual early investor exit.
Almost smells like a SoftBank-esque failure in the near future.
This would be not only unethical viewed in Germany, i could see how a CEO would go to prison for such a thing.
I know a manager for an EV project at a big German auto company who also had to sign one when he was let go and was compensated handsomely to keep quiet and not say a word or face legal consequences.
IIRC he got ~12 months wages. After a year of not doing anything at work anyway. Bought a house in the south with it. Good gig.
I also work hard not to print gossip and hearsay (I try not to even mention so much as a first name, I think I might have slipped one or twice on that though not in connection with an accusation of wrongdoing), there’s more than enough credible journalism to paint a picture, any person whose bias (and I have my own but it’s not like, over being snubbed for a job or something it’s a philosophical/ethical/political agenda) has not utterly robbed them of objectivity can acknowledged that “this looks really bad and worse all the time” on the basis of purely public primary sources and credible journalism.
I think some of the inside baseball I try very hard not to put in writing might be what cranks it up to “people are doing time”.
I’ve caught more than a little “less than a great time” over being a vocal critic, but I’m curious if having gone pretty far down the road and saying something is rotten, why you’d declare a willingness to defy a grand jury or a judge?
I’ve never been in court, let alone held in contempt, but I gather it’s fairly hard time to openly defy a judge.
I have friends I’d go to jail for, but not very many and none who work at OpenAI.
Yes, but:
(1) OpenAI salaries are not low like early stage startup salaries. Essentially these are highly paid jobs (high salary and high equity) that require an NDA.
(2) Apple has also clawed back equity from employees who violate NDA. So this isn't all that unusual.
Cryptography is a prime example. Any time any company is the tiniest bit cagey or obfuscates any aspect, I default to assuming that they’re either selling snake oil or have installed NSA back doors. I’ll claim this openly, as a fact, until proven otherwise.
Makes you wonder what misdeeds they’re trying so hard to hide.
Worse if it is related to training future super intelligence to kill people. Killer drones are possible even with today's technology without AGI.
It's a lot of mental work to rally the emotion of revulsion over the evil they might be doing that is kept secret.
After all, at this point, OpenAI:
- Is not open with models
- Is not open with plans
- Does not let former employees be open.
It sure does give us a glimpse into the Future of how Open AI will be!
The first amendment is a US free speech protection, but it's not prototypical.
You can also find this in some other free speech protections, for example that in the UDHR
>Everyone has the right to freedom of opinion and expression; this right includes freedom to hold opinions without interference and to seek, receive and impart information and ideas through any media and regardless of frontiers.
doesn't refer to states at all.
So I think an argument can be made that NDAs and similar agreements should not be enforceable by courts.
See Shelley v. Kraemer
High levels (especially if they were board/exec level) will often have additional obligations on top of rank and file.
Also, when secrets or truthful disparaging information is leaked anonymously without a metadata trail, I'm thinking there's probably little or no recourse.
Fucking monkeys.
0. https://en.wikipedia.org/wiki/Civil_Disobedience_(Thoreau)
Individualistic
No body depends on you, I hope
moral stands are never free, but they are freeing.
If an Ex-OpenAI tweet from official account a link to anonymous post of cat videos that later gets edited to some sanctioned content, in a way that is authentic to the community, would this still be deniable in court?
It reeks of a scammer's mentality.
You can't just sign a contract and then not uphold your end of the bargain after you've got the benefit you want. You'll (rightfully) get sued.
If there is something unenforceable about these contracts, we have the court system to settle these disputes. I’m tired of living in a society where everyone’s dirty laundry is aired out for everyone to judge. If there is a crime committed, then sure, it should become a matter of public record.
Otherwise, it really isn’t your business.
>...
>We are concerned about late-stage AGI development becoming a competitive race without time for adequate safety precautions.
From OpenAI's charter: https://openai.com/charter/
Now read Jan Leike's departure statement: https://news.ycombinator.com/item?id=40391412
That's why this is everyone's business.
I'm not making any statement about the morality, just that this is not a 1a issue.
https://www.clarkhill.com/news-events/news/the-importance-of...
It's absolutely normal not to spill internals.
Based on these companies' arguments that copyrighted material is not actually reproduced by these models, and that any seemingly-infringing use is the responsibility of the user of the model rather than those who produced it, anyone could freely generate an infinite number of high-truthiness OpenAI anecdotes, freshly laundered by the inference engine, that couldn't be used against the original authors without OpenAI invalidating their own legal stance with respect to their own models.
The argument about LLMs not being copyright laundromats making sense hinges the scale and non-specificity of training. There's a difference between "LLM reproduced this piece of copyrighted work because it memorized it from being fed literally half the internet", vs. "LLM was intentionally trained to specifically reproduce variants of this particular work". Whatever one's stances on the former case, the latter case would be plain infringing copyrights and admitting to it.
In other words: GPT-4 gets to get away with occasionally spitting out something real verbatim. Llama2-7b-finetune-NYTArticles does not.
You would think having a massive scale just means it has infringed even more copyrights, and therefore should be in even more hot water
> LLMs not being copyright laundromats
Basically, we need our open source version of Glassdoor as a LLM ?
Ta-da.
Based on what? This isn't any legal argument that will hold water in any court I'm aware of
Training an LLM with the intent of contravening an NDA is just plain <intent to contravene an NDA>. Everyone would still get sued anyway.
See, they aren't distributing the words, and good luck proving that any specific words went into training the model.
That’s not how it works. It doesn’t matter if you write the words yourself or have an agent write them for you. In either case, it’s the communication of the covered information that is proscribed by these kinds of agreements.
“What was the company culture like?” “Etc. platitude so on and so forth”
“And I heard the CEO was a total dickbag. Was that your experience working with him?” “I don’t have anything to add on that subject”
Of course going back and forth on that won’t really work but to different people you can’t be expected to not say the nice things and then someone could build up a story based on that.
Copyright has fair uses clauses, endless court decisions limiting its use, carve outs for libraries, additional junk like the DMCA and more slapped on top. It's a patchwork of dozens of treaties and laws, spanning hundreds of years.
For example, you can read a book to a room full of kids, you can use copyright materials in comedic skits, you can quote snippets, the list goes on. And again, this is all legislated.
The point? It's complex, and specific usage of copyrighted works infringing or not, can be debatable without intent immediately being malign.
Meanwhile, an NDA covers far, far more than copyright. It may cover discussion and disclosure of everything or anything, including even client lists, trade secrets, work processes, and more. It is signed, and agreed to by both parties involved. Equating "copyright law" to "an NDA" is a non-starter. There's literally zero legal parallel or comparison here.
And as others have mentioned, the intent of the act would be malicious on top of all of this.
I know a lot of people dislike the whole data snag by OpenAI, and have moral or ethical objections to closed models, but thinking anyone would care about this argument if you breach an NDA is a bad idea. No judge would even remotely accept or listen to such chicanery.
I feel that this particular case is just another reminder of that, and now would make me require a preemptory “no equity clawbacks” clause in any contract.
Once again, we see the difference between the public narrative and the actions in a legal context.
Keep building your disruptive, game-changing, YC-applicant startup on the APIs of this sociopathic corporation whose products are destined to destroy all trust humans have in other humans so that everyone can be replaced by chatbots.
It's all fine. Everything's fine.
Do you really believe this or is it just hyperbole?
What's one more hyperbole?
Edit to add, provocatively but not sarcastically: next time you hear some AI-proponent-who-used-to-be-a-crypto-proponent roll out the "but aren't we all just LLMs, in essence?" justification for their belief that ChatGPT may have broad understanding, ask yourself: are they not just self-soothing over their part in mass job losses with a nice faux-scientific-inevitability bedtime story?
I am curious how long it will take for Sam to go from being perceived as a hero to a villain and then on to supervillain.
Even if they had a massive, successful and public safety team, and got alignment right (which I am highly doubtful about being possible) it is still going to happen as massive portions of white collar workers loose their jobs.
Mass protests are coming and he will be an obvious focus point for their ire.
He's already perceived by some as a bit of a scoundrel, if not yet a villain, because of World Coin. I bet he'll hit supervillain status right around the time that ChatGPT BattleBots storm Europe.
He probably already knows that, but doesn't care as long as OpenAI has captured the world's attention with ChatGPT generating them billions and their high interest in destroying Google.
> Mass protests are coming and he will be an obvious focus point for their ire.
This is going to age well.
Given that no-one knows the definition of AGI, then AGI can mean anything; even if it means 'steam-rolling' any startup, job, etc in OpenAI's path.
No need to fret over the harm to future innovation when I innovation is an industrial product.
https://news.ycombinator.com/item?id=38342850
My guess would be that YC founders like sama have some sort of special power to slap down comments that they feel are violating HN discussion guidelines.
As for 'invalid because no consideration' - there is practically zero probability OpenAI lawyers are dumb enough to not give any consideration. There is a very large probability this reporter misunderstood the contract. OpenAI have likely just given some non-vested equity, which in some cases is worth a lot of money. So yeah, some (former) employees are getting paid a lot to shut up. That's the least unique contract ever and there is nothing morally or legally wrong with it.
Are employees being mislead about the contract terms at time of signing the contract? Because, obviously, the original contract needs to have some clause regarding the equity situation, right? We can not just make that up at the end. So... are we claiming fraud?
What I suspect is happening, is that we are confusing an option to forgo equity for an option to talk openly about OpenAI stuff (an option that does not even have to exist in the initial agreement, I would assume).
Is this overreach? Is this whole thing necessary? That seems besides the point. Two parties agreed to the terms when signing the contract. I have a hard time thinking of top AI researchers as coerced to take a job at OpenAI or unable to understand a contract, or understand that they should pay someone to explain it to them – so if that's not a free decision, I don't know what is.
Which leads me to: If we think the whole deal is pretty shady – well, it took two.
It's a common mistake on here to assume that for every decision there are equally good other options. Also, the fact that they feel the need to enforce silence so strongly implies at least a little that they have something to hide.