[elAhmo]

> Since then, OpenAI's board has been refreshed with more tech executives, chaired by Bret Taylor, former Salesforce co-CEO who now runs his own AI startup. Any corporate changes need approval from its nine-person non-profit board.

Why would the non-profit board approve a change to a for-profit company? Wouldn't this be against the nature of the non-profit entity that was founded and which they are supposed to govern?

[neom]

Here are the board members just to save anyone else looking it up:

Bret Taylor (Chair), Sam Altman, Adam D’Angelo, Dr. Sue Desmond-Hellmann, Retired U.S. Army General Paul M. Nakasone, Nicole Seligman, Fidji Simo, Larry Summers and Zico Kolter.

https://en.wikipedia.org/wiki/Bret_Taylor

https://en.wikipedia.org/wiki/Sam_Altman

https://en.wikipedia.org/wiki/Adam_D%27Angelo

https://en.wikipedia.org/wiki/Sue_Desmond-Hellmann

https://en.wikipedia.org/wiki/Paul_Nakasone

https://en.wikipedia.org/wiki/Nicole_Seligman

https://en.wikipedia.org/wiki/Fidji_Simo

https://en.wikipedia.org/wiki/Lawrence_Summers

https://csd.cmu.edu/people/faculty/zico-kolter

[pyinstallwoes]

> In May 2018, he became head of the National Security Agency, the Central Security Service and the United States Cyber Command.

[dirkc]

I don't think you can vote to change a non-profit to a for-profit company. Once a non-profit, always a non-profit.

There are sort-off loopholes, like changing the name of "Open AI" to something else and selling the name along with the IP to a commercial entity - EdX did this. But it's not really a loophole, since the board would need justify the price the commercial company paid and the money from the sale is still within a non-profit and bound by it's rules.

[gwern]

Exactly. The problem here is that OA is too valuable, and the non-profit owns too much of it (ie. all of it). The non-profit owns exactly 100% of the OA for-profit right now, and can cancel the "PPUs" any time it pleases, and if you believe Altman about the agent roadmap, would be required to do so within a few years. Now, given that the PPUs are being raised at a pseudo-marketcap of $150b, we're hearing, and that non-profits are required to sell assets for fair market value, how can the board sell its 100% ownership of the for-profit for anything less than the $150b that the market values that ownership at...? And where does one get $150b, exactly? Even Sam Altman can't pull that off.

So, the whole question has been, how does he figure out how to leave the board with <50% ownership (preferably much less than, even 0%), in a way which passes the legal sniff test so the deal goes through, but which doesn't sacrifice >$75b that neither he nor anyone else has?

This is where the rhetoric and preparing the grounds comes in. You can argue that OA is actually worth <<$150b, maybe even as low as $0, by saying that it has value only because it is going to keep raising enough cash to reach AGI... but the board remaining in charge + the current cap on raising capital + Altman quitting will destroy that raising ability, rendering OA worthless. (Similar to his threat last year to destroy OA by creating a competitor in MS, and MS then immediately cutting off hardware & capital so the rump OA would starve to death.)

Hence, the board should - nay, is required by their legal & moral duties to the nonprofit - to accept much less than $150b, because whatever that deal is, it's more than $0.

This is of course ridiculous and wrong, but it's not so ridiculous that a board of loyalists can't sign off on it, nor that lawyers can't defend it in court in front of a judge who wants to rule in its favor (because judges will tolerate anything in nonprofits short of gross criminality).

[gwern]

https://www.wsj.com/tech/ai/openais-complex-path-to-becoming...

[d--b]

This reeks of conflicts of interests

[sirspacey]

No conflict, no interest