[syrusakbary]

What is this article really about? So many details missing about the $70k bill to assume is the wrongdoing of the CEO or company.

The article smells like someone wants to build an argument against Stable Diffusion.

I don't think that any of the arguments on the article hold to imply that Stable Diffusion (company) is crumbling.

As an example, the "I sold 15% of the stake for $100" argument from the previous cofounder doesn't hold unless proven otherwise by a judge. The context always matter: is not the same selling 15% when there were no investors or IP, to selling it under coercion... but that a judge will be way better to assess than people external to the matter.

I wonder what's the gain behind publishing this kind of "yellow" articles (and to whom)

[mikhmha]

Really? I think the untrustworthiness of the CEO and their tendency to embellish facts around their background is a massive red flag. These have been reported on multiple times the past few months. Why lie about having a masters degree or working at a trading firm?

And of course you may say that all CEOs are liars. And I would agree. But the smart CEO’s know not to lie about trivial things like education credentials.

I don’t think this is a hit piece. I think the incompetency of the CEO is finally bubbling up to the surface and rotting the company from the inside. And now the original believers are starting to turn on their former saviour.

[cma]

> Why lie about having a masters degree or working at a trading firm?

Maybe not really having the degree but it mostly being satisfied is a positive with venture capitalist after the richest man in the world had problems with his physics degree.

[emadm]

It was a weird allegation when I paid the postage and got the MA from Oxford a couple of weeks later

https://twitter.com/emostaque/status/1682091613278072832?s=4...

Told them that and more https://emad.posthaven.com/on-setting-the-record-straight

On the cofounder share sale thing it was the Ron Wayne effect with massive sour grapes and him straight up lying

https://twitter.com/emostaque/status/1680774535342358528?s=4...

[ilaksh]

Can you detail the years of study after your completion of the BA that enabled you to be awarded the MA? Or is this a "free" MA indicating "seniority"?

If it's the latter, then that's deliberately misleading to keep referencing it as if it were a normal MA. I do appreciate your efforts with the open source models though.

[emadm]

Its the standard nomenclature to use when you have graduated from Oxford with 7 years post matriculation https://en.wikipedia.org/wiki/Master_of_Arts_(Oxford,_Cambri...

I don't get the issue when this is a standard thing, was actually at Oxford same time as Mustafa Suleyman who dropped out and is doing just fine, as is Sam at OpenAI who also dropped out of Stanford.

So its not a qualification thing, nor is it a deliberate thing to mislead when that's how you're meant to identify your degree (still weird maths and computer science is an arts degree tbh there).

[ilaksh]

"7 years post matriculation" -- meaning _only_ that seven years passed since you graduated.

The issue is that describing it as an MA implies that it is equivalent to a normal Master of Arts, when in fact you did not do any further work beyond the BA. Other Oxford graduates may "standardly" benefit from this "confusion", but it's not fair to all of the millions who have worked hard for years to achieve a real graduate degree.

Whether some other famous people dropped out of an undergraduate or graduate degree program or not does not mean that it is okay to pretend that you have completed a master's degree program of work. I do believe qualifications like that are given more weight than warranted, but that doesn't mean that it's okay to be deliberately misleading. Even having lots of other people from the same university doing it doesn't make it okay.

I didn't complete any degree at all. I don't go around telling people (in so many words) that I attended UCSD. Even though I did. Because just saying that without clarification would imply that I graduated.

[anonylizard]

There is 100% a hit piece effort against stability.

It is extremely unusual, for multiple media outlets to repeatedly write an article against a small startup, that's completely unrelated to customer harm or say employee abuse. Wow, the founder fudged his VC presentations, wow, the masters degree wasn't legit, like WHO CARES?

The only missing piece was the motivation, who exactly holds such a grudge against Emad really?

Then the missing piece finally came to light: https://www.reuters.com/technology/stability-ai-is-sued-by-c...

Basically, this "Cyrus Hodes" guy, thought stability was worthless before StableDiffusion, sold out his stake for $100. That is a fair assessment, because Stability before SD was completely aimless and going nowhere, and definitely not worth anything.

However, Emad spotted an incredible bet, and obtained the naming rights to a latent diffusion model for cheap. And the titanic success of SD instantly propelled Stability to a serious player on the AI world. Hence the 100 mil raise afterwards.

Needless to say, this Cyrus guy is extremely mad. If you check his credentials, he's not a technical guy, but actually extremely political, full of adviser roles to OECD or governmental organizations. Therefore its unsurprising that he:

1. Is able to organize the multi-month media effort in secret, to give journalists the angle to write against Stability.

2. Is ultra bitter about being kicked out of stability (But I doubt he would have been a good leader anyways at Stability, since AI transitioned from being a hype-based industry to a real product driven one after SD.)

In any case, Stability has had some hiccups, but they appear to be recovering very well in the recent months. They've had 3 separate product launches.

1. SDXL, the first successor to SD that finally has community acceptance (So the actual useful fine-tunes and ecosystems can grow around it)

2. Stablebeluga, a fast-follower high quality fine-tune for Llama2 with decent acceptance

3. Stablecode, a open source coding-specific LLM

It shows their R&D team is once again working, and gaining the confidence of the community again, after the two disasters that was SD2.0 and stableLM

[mikhmha]

I always see this line of reasoning online from hype consumers. The eternal victim complex. Its not fair that negative articles are being written about this humble startup/video-game merchant/crypto exchange - must be a coordinated hit piece!

And then it devolves into worshipping the founder/priest/oracle. Usually by making claims of their supposed genius and foresight. Blah so boring. You can see it all on display in this comment. Emad the genius, yet he lies about trivial things. Yeah who cares that he lied about his master's degree? I mean I agree, who gives a shit about a master's degree. But then why feel the need to lie about it? You don't think that reflects on his character?

[ShamelessC]

> Needless to say, this Cyrus guy is extremely mad. If you check his credentials, he's not a technical guy, but actually extremely political, full of adviser roles to OECD or governmental organizations. Therefore its unsurprising that he:

> 1. Is able to organize the multi-month media effort in secret, to give journalists the angle to write against Stability.

This is cryptocurrency levels of conspiracy theory. Do you have any evidence? Or do you have an existing track record to back up these claims at least?

[predictabl3]

> Wow, the founder fudged his VC presentations, wow, the masters degree wasn't legit, like WHO CARES

I gotta stop coming here. I just have to. Board the hype train and proudly ignore the red flags.

Choo choo, or something. I swear to God reading the comments here, on average, is making me dumber lately.

Can we fork the community? The MBA, PM, hype beast clout chasers can go to one side, and the folks pushing to GH 10 times a day and actually building stuff can have another?

[the__prestige]

It feels like this reply is posted by Emad Mostaque

[dontknowmuch]

Looks like a hit piece planted to build momentum to oust the CEO.

[tmaly]

I personally think the Stable Diffusion product being somewhat open in that I can download it and run it on my home desktop is pretty amazing. That along is encouraging some much interesting innovation in this new type of art.

People are creating all sorts of cool LoRas.

[formvoltron]

why would anyone bother to sell for $100? I mean just get a job as a UPS driver!

[defen]

It's possible that he perceived his ownership stake as having negative value - e.g. due to tax reasons or other regulatory overhead (needing to participate in board meetings or whatever), legal risk (they get sued for doing something desperate and dumb and he needs to participate in the lawsuit) etc. But you can't just dissolve your shares, you have to either shut the company down entirely or sell your stake to someone. If he thought the company was truly worthless / negative expected value then $100 is fair consideration.

A famous example of this is Ronald Wayne selling his 10% stake in Apple for $800 because he was worried that Jobs would incur a bunch of risky debt that he'd be liable for.

[chii]

i would've imagined that a limited liability corporation means that the lowest value it can go is zero, not negative.

A shareholder cannot and should not ever be responsible for the debts of the company imho. As for time costs, the shareholder could just not attend or do anything, if they perceive the work to be of worthless value, and thus, should therefore have zero as the floor, rather than negative!

[esperent]

In case anyone is out of the loop here, UPS drivers have a strong union and negotiated a great rate of pay. The story being spun in the media is that us tech workers are angry about this for some reason.

It's one of the most blatant attempts to pit workers against workers and undermine unions that I've ever seen.

[NegativeK]

I've got some incredibly worthless stuff that I'd take $100 for.

[sokoloff]

Is any of it a substantial stake in an operating generative-AI company? If so, I'll buy it for $100; I literally don't intend to do more diligence than that to find a likely overlay.

[emadm]

He wanted out completely and no association after first project failed

https://twitter.com/emostaque/status/1680774535342358528?s=4...

[sokoloff]

According to the account of someone who has a substantial vested interest in that being the case. (It might be the case, in which "they got what they asked for".)

In a case where each side has a substantial vested interest in their story being accepted as true, it's wise to be skeptical of the claims on both sides and look for other, corroborating or contradicting evidence.

[emadm]

It's pretty clear cut as his suit has three claims

1. Did not inform him of pivot and generative AI art (false, even generated AI art for him)

2. Did not inform him of fundraise/inflows (also false)

3. Either of the above are not legal under Delaware law (also false)

We will find out in a few months I suppose.

[nonbirithm]

I don't know if this is 100% a hit piece if so many people really did resign citing emad as the main problem. If he is a compulsive liar or as incompetent as his co-workers claim then it's a red flag to me.

I remember the one time talked about how he could scrape data behind paywalls and about deliberately bypassing anti-scraping techniques on Twitter. I don't think it was a very good PR move.

In the back of my mind I always had this thought that Stability could be some kind of AI accelerationist fund - hype up a shining future, raise enough capital to clear the bar for model training, release their model unencumbered and gather a fervent following of gracious users, and... not have a fully realized plan for what comes after.

But I have a feeling that even if they go under people will still be talking about Stability and their influence on generative art for a long while, at least until another freely available model can oust Stable Diffusion.

[emadm]

That was just the hate mob attacking

I noted that the most interesting data is behind firewalls as in private data as our entire model is open models to private data => not using that data to train our models

You can also license that data smartly, with royalty schemes and more, three no reason to get past paywalls etc nor do we do so (others do of course)

We are the only AI company to even offer opt outs, first models soon on that