Yesterday, Sam Altman and Greg Brockman were fired from the Board of Directors of OpenAI. Following, all of Tech Twitter was abuzz with one question: wait a moment, who was on the Board? And after they found out, they asked: who on earth are Tasha McCauley
- The Open Phil seat deal was only for 3 years, and has long since expired, so while Karnofsky may have nominated Toner, he couldn't've forced her in nor can Toner pass it on to a person of her choice, so when Altman began saying she should be fired for criticizing OA, the seat was up for grabs. (AFAICT, OpenPhil has been a bystander to all this.)
- I haven't come across any evidence that Hurd's departure is other than it seems: political campaigns are short & brutal, so the fact that he dropped out a few months later means simply that his campaign didn't take off but crashed & burned. Hurd's departure is pretty much forced - one of the only red lines for a 501(c)3 like OA nonprofit is touching a political campaign or specific elected official: https://en.wikipedia.org/wiki/501(c)(3)_organization#Limitations_on_political_activity So as soon as he decided to run, he needed to leave.
Thanks for the thoughtful comment! I came across some of the same sources that you did, and have some more comments to add:
1. It appears uncertain that the Open Phil seat deal was for 3 years. While the donation was $10M a year for 3 years, there's no evidence that the seat was limited. Most tellingly, the grant was made in March 2017. If it was a three-year-deal, then the seat would've expired in March 2020. But Holden Karnofsky remained on the board for another year-and-a-half, into September 2021. My read is that the board seat was in perpetuity.
2. On both Reid and Zilis, for me there isn't enough to go on to interpret their actions beyond what I wrote in the article.
3. Agreed on the stalemate.
4. Thanks for the note on Hurd's departure, I wasn't familiar with the restrictions on political activity for 501(c)3s.
1. They haven't commented much about the details publicly, but an OPer has told me that it was only for 3 years and it had in fact expired before Holden left. So I'm quite confident.
4. Yeah, it's not that well-known but you can tell how seriously they take it by how rarely you ever spot a 501c3 talking about a specific candidate or election, despite there being so many extremely political 501c3s of every variety. They're careful to keep it to general issues, background research, funding individuals with known politics, or PACs or other charitable organizations which aren't 501c3s and so allowed to advocate like that. (I was, for example, looking the other day into a liberal nonprofit which paid a guy for years to maliciously edit Wikipedia with a stable of sockuppets and ban-evading etc, and AFAICT, the only thing about this which *might* have been illegal on the nonprofit's part is if he had been foolish enough to edit articles about some of the politicians involved in their particular area of lobbying, because they are a 501c3 and not one of the others: https://www.reddit.com/r/media_criticism/comments/17zqwgb/the_hofmann_wobble_by_ben_lerner/kagx946/ )
It should be noted the primary financier of Open Philanthropy is multi-billionaire co-founder of Facebook, Dustin Moskovitz, who also participated in Angel and Series A rounds of Anthropic.
Dustin may be a true believer in AI safety and Effective Altruism ideology as is the stated mission of Open Philanthropy, but the skeptic in me would not rule out him using those public positions to cynically further his own private objectives, whatever they may be, as is common among many billionaire philanthropists. If anything, it's worth a further look.
It's intriguing how Adam, the founder of Quora, has essentially been building the Poe chatbot for Quora, which leverages OpenAI's APIs. All this time, there was an impression that large language models had utilized all the Quora search answers. However, with Adam on the board and Sam being an investor in Quora since 2018, the spotlight is on the board dynamics of a non-profit company (apparently).
Some additions:
- Hoffman didn't resign voluntarily, Altman forced him out: https://www.semafor.com/article/11/19/2023/reid-hoffman-was-privately-unhappy-about-leaving-openais-board
- Shivon Zilis stepped down not over the twins, which was old news, but only after Musk broke Twitter's contract with OA and began heavily criticizing it: https://www.semafor.com/article/03/24/2023/the-secret-history-of-elon-musk-sam-altman-and-openai
- The Open Phil seat deal was only for 3 years, and has long since expired, so while Karnofsky may have nominated Toner, he couldn't've forced her in nor can Toner pass it on to a person of her choice, so when Altman began saying she should be fired for criticizing OA, the seat was up for grabs. (AFAICT, OpenPhil has been a bystander to all this.)
- The lack of board nominations was due to a stalemate: see https://www.nytimes.com/2023/11/21/technology/openai-altman-board-fight.html https://www.wsj.com/tech/ai/altman-firing-openai-520a3a8c
- I haven't come across any evidence that Hurd's departure is other than it seems: political campaigns are short & brutal, so the fact that he dropped out a few months later means simply that his campaign didn't take off but crashed & burned. Hurd's departure is pretty much forced - one of the only red lines for a 501(c)3 like OA nonprofit is touching a political campaign or specific elected official: https://en.wikipedia.org/wiki/501(c)(3)_organization#Limitations_on_political_activity So as soon as he decided to run, he needed to leave.
Thanks for the thoughtful comment! I came across some of the same sources that you did, and have some more comments to add:
1. It appears uncertain that the Open Phil seat deal was for 3 years. While the donation was $10M a year for 3 years, there's no evidence that the seat was limited. Most tellingly, the grant was made in March 2017. If it was a three-year-deal, then the seat would've expired in March 2020. But Holden Karnofsky remained on the board for another year-and-a-half, into September 2021. My read is that the board seat was in perpetuity.
2. On both Reid and Zilis, for me there isn't enough to go on to interpret their actions beyond what I wrote in the article.
3. Agreed on the stalemate.
4. Thanks for the note on Hurd's departure, I wasn't familiar with the restrictions on political activity for 501(c)3s.
1. They haven't commented much about the details publicly, but an OPer has told me that it was only for 3 years and it had in fact expired before Holden left. So I'm quite confident.
(The time-limited three-year is also implied by their discussion on their website which talks about it being tied to the three-year grant which they did not renew https://www.openphilanthropy.org/grants/openai-general-support/ , and further implied by Moskowitz recently talking about it in the distant past tense when asked about it like on https://www.lesswrong.com/posts/Qz5DGNKeX9v8GRCnL/what-happened-to-the-openphil-openai-board-seat#comments )
4. Yeah, it's not that well-known but you can tell how seriously they take it by how rarely you ever spot a 501c3 talking about a specific candidate or election, despite there being so many extremely political 501c3s of every variety. They're careful to keep it to general issues, background research, funding individuals with known politics, or PACs or other charitable organizations which aren't 501c3s and so allowed to advocate like that. (I was, for example, looking the other day into a liberal nonprofit which paid a guy for years to maliciously edit Wikipedia with a stable of sockuppets and ban-evading etc, and AFAICT, the only thing about this which *might* have been illegal on the nonprofit's part is if he had been foolish enough to edit articles about some of the politicians involved in their particular area of lobbying, because they are a 501c3 and not one of the others: https://www.reddit.com/r/media_criticism/comments/17zqwgb/the_hofmann_wobble_by_ben_lerner/kagx946/ )
Who defines "highly ideological AI governance organization"? I don't think it's nearly that simple.
This is incredibly informative!
It should be noted the primary financier of Open Philanthropy is multi-billionaire co-founder of Facebook, Dustin Moskovitz, who also participated in Angel and Series A rounds of Anthropic.
Dustin may be a true believer in AI safety and Effective Altruism ideology as is the stated mission of Open Philanthropy, but the skeptic in me would not rule out him using those public positions to cynically further his own private objectives, whatever they may be, as is common among many billionaire philanthropists. If anything, it's worth a further look.
https://www.politico.com/news/2023/10/13/open-philanthropy-funding-ai-policy-00121362
It's intriguing how Adam, the founder of Quora, has essentially been building the Poe chatbot for Quora, which leverages OpenAI's APIs. All this time, there was an impression that large language models had utilized all the Quora search answers. However, with Adam on the board and Sam being an investor in Quora since 2018, the spotlight is on the board dynamics of a non-profit company (apparently).