内容来源：https://www.theverge.com/ai-artificial-intelligence/920775/evidence-exhibits-elon-musk-sam-altman-openai-trial

内容总结：

马斯克诉奥尔特曼案庭审：OpenAI早期内幕邮件曝光

加州联邦法院本周一正式开庭审理马斯克诉奥尔特曼案，随着案件推进，大量早期邮件、照片和公司文件陆续公开，揭示了OpenAI从创立之初到转向营利模式期间的内部博弈与理念冲突。

案件核心：OpenAI是否背离初心？

马斯克起诉奥尔特曼、布罗克曼以及微软，指控被告违反慈善信托、欺诈及不当得利。案件焦点在于OpenAI是否背离了其创始使命——确保通用人工智能（AGI）惠及全人类。作为OpenAI联合创始人及早期投资者，马斯克此前已多次对该公司及其高管采取法律行动。

关键人物与早期分歧

最新披露的文件显示，英伟达CEO黄仁勋曾向OpenAI提供一台抢手的超级计算机，马斯克亲自起草了OpenAI的大部分使命宣言并深刻影响了其早期架构。然而，OpenAI总裁布罗克曼和首席科学家苏茨克维对马斯克的控制权表示担忧。

一封2017年8月的邮件显示，马斯克的首席幕僚齐利斯在汇报与布罗克曼、苏茨克维的会议时指出，双方最大的矛盾点在于马斯克对公司的控制期限——尽管他拥有股权。两位创始人坚持“铁律”：任何个人都不能对AGI拥有绝对控制权，哪怕创始团队发生变故，权力也必须在2-3年后分散。

马斯克的回复直截了当：“这很烦人。请鼓励他们自己去开公司。我受够了。”

博弈升级：控制权之争白热化

2017年9月的一封邮件中，布罗克曼和苏茨克维直接向马斯克表达了担忧：“当前的结构为你提供了对AGI的单方面绝对控制路径……你曾说不想控制最终的AGI，但在谈判中你向我们表明，绝对控制对你极其重要。”

二人还质疑奥尔特曼的动机：“我们一直无法完全信任你的判断……CEO头衔为什么对你如此重要？你的说法一直在变。”奥尔特曼回应称“仍对非营利结构充满热情！”

对此，马斯克发出最后通牒：“我受够了。这是最后一根稻草。要么你们自己干，要么继续以非营利形式运营OpenAI。在你们做出坚定承诺之前，我不会再资助OpenAI。”

转向营利：理念与现实的冲突

2018年初，马斯克在邮件中警告OpenAI“相对于谷歌注定失败”，认为“必须立即采取重大行动”。当时从OpenAI跳槽至特斯拉的卡帕西建议：最好的方案是让OpenAI依附于特斯拉作为“现金牛”，利用特斯拉的供应链和车载计算平台，通过自动驾驶解决方案创造收入，再反哺AI研发。

马斯克将卡帕西的分析转发给苏茨克维和布罗克曼，写道：“特斯拉是唯一有可能与谷歌抗衡的路径。即便如此，概率也很小，但至少不是零。”

分手与后续

2018年2月，马斯克通过视频会议宣布退出OpenAI董事会。齐利斯在短信中报告苏茨克维“明显崩溃”，并询问是否要挖角。马斯克回应：“如果我专注特斯拉AI，OpenAI成为重要力量的可能性微乎其微。”

同年8月，奥尔特曼向马斯克发送了营利性有限合伙企业的条款清单，明确标注“投资OpenAI LP属于高风险投资”，并声明“我们对使命的忠诚优先于盈利义务”。奥尔特曼表示自己将不持有任何股权。

2020年9月，马斯克公开批评微软获得GPT-3独家授权“有违开源精神，OpenAI实际上已被微软俘获”。奥尔特曼辩称：“没有数十亿美元我们无法与DeepMind抗衡，微软是以最小妥协获取资金的最佳途径。”

庭审仍在继续

随着陪审团审判的推进，预计将有更多证据浮出水面。该案结果可能直接影响OpenAI的商业模式及其对快速发展的AI技术的控制方式，而OpenAI与SpaceX今年均计划上市，使得这场法律战格外引人关注。

中文翻译：

马斯克诉奥特曼案庭审正在进行中，这意味着呈堂证供正逐步公之于众。截至目前，OpenAI创立初期——甚至在该人工智能实验室尚未命名之前——的电子邮件往来、照片及公司文件正在流传。以下是一些关键要点：英伟达CEO黄仁勋向OpenAI提供了一台抢手的超级计算机；马斯克主要起草了OpenAI的使命宣言，并对其早期结构产生了重大影响；OpenAI CEO萨姆·奥特曼似乎希望大力依靠Y Combinator为OpenAI提供早期支持；OpenAI总裁格雷格·布罗克曼和伊利亚·苏茨克维对马斯克对公司的控制程度感到担忧；马斯克则强调了以广泛有益的人工智能为使命的非营利组织的重要性。

马斯克诉奥特曼案迄今已披露的全部证据

追溯到2015年的电子邮件让人得以一窥OpenAI的基础及其内部的早期紧张局势。

马斯克备受关注的诉讼于周一在加州联邦法院开始陪审团审判，被告包括奥特曼、布罗克曼以及OpenAI的投资方微软。对每一方的指控各不相同，包括违反OpenAI的慈善信托、欺诈以及不当得利。但归根结底，马斯克的诉讼归结于OpenAI是否偏离了其创始使命——确保通用人工智能（一个通常定义模糊的术语，指代达到或超越人类智能的人工智能系统）造福全人类。这是马斯克多年来对OpenAI及其高管提起的一系列法律行动中的最新一起。马斯克与奥特曼和布罗克曼共同创立了该人工智能实验室，并曾是早期投资者。（马斯克还拥有xAI，这是一家与OpenAI直接竞争的人工智能实验室，归母公司SpaceX所有。）

前OpenAI员工以及熟悉两家公司的人士一直在密切关注这起特定的诉讼，因为陪审团审判的结果可能会影响OpenAI如何运营其业务以及控制其快速发展的技术。此外，据报道，OpenAI和SpaceX都力争在今年上市，因此它们比以往任何时候都更受公众关注。

诉讼证据开示程序已经挖出了人工智能行业高管之间大量令人侧目的通信，从奥特曼与苏茨克维之间的电子邮件，到布罗克曼自己的日记条目。甚至Meta CEO马克·扎克伯格与马斯克之间的短信也被公开。但这都是在陪审团审判开始之前——现在，还有更多证据即将披露。

以下是迄今为止所有已公开证据的详尽列表，以及每项证据最重要的启示。诚然，并非每一条都一定有趣，因此我们用星号标记了最重要的部分。《The Verge》将随着更多证据的加入持续更新此列表。

2026年4月29日公布的文件

• 2015年6月奥特曼与马斯克之间的电子邮件往来。 奥特曼提出了一项包含五个部分的计划，涉及一个人工智能实验室，其使命是“创造第一个通用人工智能，并用于赋予个人权力——即，看似最安全的分布式未来版本。更广泛地说，安全应成为首要要求。”
  他建议从七到十个人开始，然后逐步扩大，使用位于山景城的额外Y Combinator大楼。在治理方面，奥特曼最初提名了五个人，提议自己、马斯克、比尔·盖茨、皮埃尔·奥米迪亚和达斯汀·莫斯科维茨。“该技术将由基金会拥有，并‘为了世界的利益’而使用，在如何应用不明确的情况下，将由我们五个人来决定，”奥特曼写道。他补充说，在实验室工作的研究人员将拥有“显著的经济收益……与他们所创造的东西无关，这应能消除一些冲突”，并建议向他们支付“具有竞争力的薪水”并授予Y Combinator的股权。他还说，应该找一个人来“管理团队”，但那个人“可能不应该进入治理委员会”。
  奥特曼接着问马斯克，除了治理之外，他是否会参与人工智能实验室，可能会每月来一次讨论进展，或者至少在公开场合表示支持以帮助招聘。他以彼得·蒂尔在Y Combinator的“兼职合伙人”身份为例。
  最后，奥特曼提到了“监管信函”，似乎暗示该人工智能实验室将撰写一封呼吁人工智能监管的信函。他说他很乐意将马斯克从公开签署人中排除。
  马斯克回复：“全部同意。”

• 2015年10月奥特曼与马斯克之间的电子邮件往来。 奥特曼建议从马斯克承诺的1亿美元开始，并问他能否在未来五年内额外捐赠3000万美元。他说比尔·盖茨尚未承诺捐赠，但他希望“下周能把他敲定”，并补充说他认为马克·扎克伯格可能不会参与，因为他自己有人工智能实验室Facebook AI Research (FAIR)。他还建议他和马斯克先作为安全委员会的头两名成员，并有可能在接下来的一年内增加另外三名成员，称其为“发布任何可能有危险东西的‘第二把钥匙’”。
  马斯克回应：“我们谈谈治理问题。这至关重要。我不想资助一个最终走向错误方向的东西。”

• 2015年11月马斯克与奥特曼之间的电子邮件往来。 两人讨论即将成立的人工智能实验室的计划。马斯克首先提到与格雷格·布罗克曼进行了一次“很棒的谈话”，并说“到目前为止，我对每个人都印象深刻”，称其为“伟大的团队”。他建议将该实验室创建为一个“独立的、纯粹的501c3组织，但明确专注于将强大的通用人工智能广泛传播给人类的积极前景”。他说，该公司“仍然旨在某个时间点实现收入超过成本，但正净收入将直接流入现金储备。”
  关于员工薪酬，马斯克建议提供现金工资和某些奖金。他说，如果奥特曼同意，员工可以将现金转换为Y Combinator的股票，并补充说，如果他们更愿意将部分或全部转换为SpaceX股票，也可以。“在SpaceX这边，因为它是私有的，我几乎可以为所欲为（谢天谢地），”马斯克写道。他还提出提供来自特斯拉的“海量真实世界传感器数据”供人工智能实验室使用，并提到数据量“比任何其他公司都高出几个数量级”。
  马斯克为人工智能实验室起的第一个名字是“Freemind”，他说这“传达了一种感觉，即我们正试图创造一种所有人都能自由获取的数字智能——这与Deepmind的‘一统天下’方法相反。”他还说，他会投入任何有用时间，即使这可能意味着分配给SpaceX和特斯拉的时间减少。“如果我真正相信这可能是近期内最大的生存威胁，那么行动应该跟上信念，”他写道。他后来补充说，尽管他似乎试图充当一个沉默的合伙人，但他必须“硬着头皮承认实际参与。这对许多人来说会是一个震惊，但就这样吧。对此不能半心半意。”
  奥特曼建议人工智能实验室与Y Combinator共用一栋大楼，并利用该孵化器的法律团队来帮助起步。他还建议使用“Axon”或与著名计算机科学家和数学家艾伦·图灵相关的名称。
  马斯克写道：“与图灵相关、听起来不那么不祥的东西可能不错。不过要避免与图灵测试联系起来，因为那听起来太像我们正在取代人类了。”

• 2015年12月奥特曼与马斯克之间的电子邮件往来。 两人起草了OpenAI使命宣言和新闻稿的开头段落。马斯克说，“这份新闻稿的全部意义在于吸引顶尖人才。”两人就措辞反复推敲，最终成品与马斯克的初稿相差不大。
  马斯克在草稿中写道：“这个事业的结果不确定，与其他公司提供的待遇相比，薪酬也不高，但我们相信目标和结构是正确的。”奥特曼在草稿中写道：“因为我们没有任何财务义务，所以我们可以专注于对人类最大的积极影响，并尽可能广泛地传播人工智能技术。”

• OpenAI的正式公司章程，于2015年12月8日提交。 文件声明OpenAI“应是一家专为慈善目的而组织的非营利公司”，其目的是“确保通用人工智能造福全人类，包括通过进行和/或资助人工智能研究。公司还可以研究和/或以其他方式支持安全开发并分发此类技术及其相关利益的努力，包括分析技术的社会影响，并支持相关的教育、经济和安全政策研究及倡议。”
  文件继续写道：“由此产生的技术将使公众受益，公司将在适用时寻求为公共利益而分发它。公司不是为了任何个人的私利而组织的。”

• 2016年4月马斯克与英伟达CEO黄仁勋之间的电子邮件往来。 马斯克问黄仁勋，OpenAI团队能否购买一台早期型号的超级计算机，并特意强调“OpenAI与特斯拉无关。它是一个由我和其他几个人资助的非营利组织，目标是开发安全的通用人工智能（并希望不是好心铺就通往地狱之路）。”
  黄仁勋回应说，他将“确保OpenAI能拿到首批之一”。

• 一张黄仁勋送来上述计算机的照片。 埃隆·马斯克站在旁边。
  他身后的墙上有一段很长的引文，有时被认为是美国海军上将海曼·G·里科弗所说，这段引文在奥特曼2013年的一篇博客文章中也有体现。（The Verge无法立即确认整段引文都是里科弗所说；在一篇属名该海军上将的美国海军帖子中，只出现了部分引文：“人类有巨大的努力能力。但它比我们想象的要大得多，以至于很少有人能达到这种能力。”）

• 2017年8月马斯克与希冯·齐里斯之间的电子邮件往来。 齐里斯是马斯克的幕僚长，后来进入OpenAI董事会，并与马斯克共同育有多个孩子。齐里斯写了一份她与布罗克曼和苏茨克维会面的总结，列出了七个悬而未决的问题。她说，布罗克曼和苏茨克维对于马斯克在公司投入更少时间、拥有更少控制权，或者投入更多时间、拥有更多控制权都可以接受，但不能接受投入更少时间却拥有更多控制权。他们还希望启动资金能远超1亿美元，因为他们担心仅所需的数据中心就要花那么多钱。她说，布罗克曼相对坚持股权平均分配。她还写道，他们担心马斯克对公司的控制。在她的笔记中，在总结他们的担忧时，齐里斯写道：“是否要求绝对控制？他们想知道是否存在这样一种情况，如果其他所有人都不同意发展方向，是否可以有一种创造性的否决权。”
  最大的紧张点，齐里斯写道，似乎在于马斯克对公司的控制期限，尽管他拥有所有权股份。“*不可协商的一点似乎是，要有一项铁定的协议，确保在创造出通用人工智能后，没有任何个人能对其进行绝对控制。满足这一点意味着要确保，无论他们三人（格雷格、伊利亚和萨姆）发生什么情况，在最初的两三年期限后，对公司的权力都会被分散……一项铁定的少数股权控制协议，期限为2-3年，无论格雷格、萨姆、伊利亚的命运如何。”
  马斯克回应：“这非常烦人。请鼓励他们自己去开公司。我受够了。”

• 2017年9月贾里德·伯查尔发给马斯克的电子邮件。 伯查尔是马斯克的顾问及家族办公室经理。他附上了一份“伊利亚和格雷格提议的资本化表，更用户友好的版本。”
  其中，马斯克的股权比例显示为51.20%，奥特曼、苏茨克维和布罗克曼各为11.01%。还有为员工预留的股权，资本化表注明了每位初始员工的名字或昵称，后跟提议的股权数量。

2026年4月30日公布的文件

• 2015年11月马斯克与奥特曼之间的电子邮件往来。 奥特曼提到了似乎是该人工智能实验室最初考虑的名称和结构之一——Y Combinator AI。
  奥特曼写道：“计划是你、我和伊利亚进入YC AI的董事会，这将是一家特拉华州的非营利组织。”他补充说，“我们会在章程中写明，任何可能危及人类安全的技术，在发布前必须获得董事会的同意，并且我们会在研究人员的雇佣合同中引用这一点。”
  马斯克在回复中不同意：“我认为这应该独立于YC（但得到其支持），而不是听起来像是一个子公司。而且，这个结构看起来也不是最优的。特别是，YC的股票加上非营利组织的薪水混淆了激励的一致性。也许建立一个标准的C型公司，同时设立一个平行的非营利组织会更好。”

• 2016年12月马斯克与其Neuralink同事之间的电子邮件往来。 他再次提到了对超越谷歌Deepmind的担忧，写道：“Deepmind行动非常迅速。我担心OpenAI没有走上迎头赶上的道路。事后看来，将其设立为非营利组织可能是一个错误的举动。紧迫感不够强。”

• 2017年6月，马斯克写邮件说他从OpenAI挖来了安德烈·卡帕斯担任特斯拉视觉部门的负责人。 他说：“OpenAI的那些家伙会想杀了我，但这必须得做……”

• 2017年7月，马斯克在给苏茨克维和布罗克曼的邮件中写道，中国“会不惜一切代价获取我们开发的东西。这也许是改变路线的另一个理由。” 布罗克曼表示同意，并说未来的道路应该是“人工智能研究非营利组织（至2017年底），人工智能研究和硬件营利组织（从2018年开始），[以及]政府项目（时间：？？）。”

• 作为对他们在OpenAI工作的感谢，马斯克提出给苏茨克维、布罗克曼和团队其他成员提供“不向公众发售”的特斯拉Model 3汽车。

• 2017年8月，马斯克询问奥特曼、苏茨克维和布罗克曼能否会面讨论OpenAI的“下一步” ——并主动提出使用“他刚在旧金山附近买的鬼屋”，尽管那里“有点疯狂和古怪，还会有派对狂欢。”

• 2017年8月下旬，马斯克与其资金经理伯查尔之间的电子邮件往来。 伯查尔写道，目前他“暂缓”向OpenAI提供马斯克通常每季度500万美元的捐款，并询问他是否应该继续暂缓。马斯克给予了肯定的答复。

• 2017年9月，马斯克、布罗克曼和苏茨克维之间的电子邮件往来。 苏茨克维建议马斯克拥有三个董事会席位，布罗克曼、苏茨克维和奥特曼各有一个。马斯克回应说，他认为自己有权任命四个董事会席位，并随后对另外三人表示了赞赏。
  马斯克写道：“我不期望立刻任命这[四个董事会席位]，但是，就像我说的，我毫无疑问会拥有公司的初始控制权，但这很快就会改变。粗略的目标是建立一个12人的董事会（如果这个董事会最终真的决定世界命运，可能需要16人），每位董事会成员都需对技术有深刻理解，至少对人工智能有基本了解，并拥有强大且明智的道德观。”

• 2017年9月，布罗克曼与马斯克之间的电子邮件往来，抄送了奥特曼和苏茨克维。 布罗克曼和苏茨克维提交了一份资本化表供马斯克批准，布罗克曼指出，他和奥特曼能投资的钱比苏茨克维多得多，但苏茨克维如果从奥特曼和/或布罗克曼那里获得以其所持股票为担保的贷款，可以投资超过250万美元。
  马斯克回复：“各位，你们逼得太紧了。我不同意。”

• 2017年9月，马斯克发给齐里斯和其他人的短信。 马斯克写道：“我们应该着手创建OpenAI B型公司（公益公司），就像我向格雷格和伊利亚承诺的那样。今晚讨论一下。顺便说一句，萨姆·奥特曼那边仍然没有消息。”

• 2017年9月，奥特曼、马斯克、齐里斯、布罗克曼、苏茨克维及马斯克幕僚长萨姆·特勒之间的电子邮件往来。 这描绘了一场高度紧张的双边谈判，马斯克和奥特曼基本上是一方，布罗克曼和苏茨克维是另一方。
  布罗克曼和苏茨克维给埃隆写道：“埃隆：我们真的很想和你共事……我们与你共事的愿望如此强烈，以至于我们很乐意放弃股权、个人控制权，让自己容易被解雇——只要能与你共事，我们不惜一切代价。”然而，他们写道，他们对马斯克控制OpenAI未来可能推出的技术感到担忧。
  “当前的结构为你提供了一条最终能单方面绝对控制通用人工智能的路径，”两人对马斯克写道。“你曾表示你不想控制最终的通用人工智能，但在这场谈判中，你向我们展示了绝对控制对你来说极其重要。例如，你说你需要成为新公司的CEO，这样每个人才会知道你是负责人，尽管你也说过你讨厌做CEO，宁愿不做。因此，我们担心，随着公司在通用人工智能方面取得真正进展，尽管你目前意图相反，但你仍会选择保留对公司的绝对控制。我们不同意你的说法，即我们离开的能力是我们最大的权力，因为一旦公司真正走上了通往通用人工智能的轨道，公司将比任何个人都重要得多。”
  两人还提到了团队经常提及的对Deepmind的德米斯·哈萨比斯的恐惧。他们对马斯克写道：“OpenAI的目标是让未来变得美好，并避免出现通用人工智能独裁。你担心德米斯可能会建立通用人工智能独裁。我们也担心。所以，创建一个如果你愿意就可能成为独裁者的结构是一个坏主意，特别是当我们可以创造其他避免这种可能性的结构时。”
  不过，布罗克曼和苏茨克维对奥特曼本人也有不同的担忧。
  在发给奥特曼的信息部分，他们写道：“在整个过程中，我们一直未能完全信任你的判断，因为我们不了解你的成本函数。我们不明白CEO头衔为什么对你如此重要。你陈述的理由一直在变，很难真正理解是什么在驱动你。”此外，他们还质疑奥特曼的部分动机，问他：“通用人工智能真的是你的首要动机吗？它如何与你的政治目标联系起来？你的思考过程是如何随着时间的推移而改变的？”
  奥特曼回复邮件称，他“仍然对非营利结构充满热情！”

• 2017年9月，马斯克对布罗克曼和苏茨克维上述担忧的回应。 马斯克写道：“各位，我受够了。这是最后一根稻草。要么你们自己去做点什么，要么继续作为非营利组织运营OpenAI。在你们做出坚定的承诺留下来，或者我继续像个傻瓜一样为你们创建初创公司提供免费资金之前，我不会再资助OpenAI了。讨论到此为止。”

• 2017年9月，齐里斯与马斯克之间的电子邮件往来。 齐里斯转述了奥特曼的一些感受，比如奥特曼在谈判中对布罗克曼和苏茨克维“失去了很多信任”，觉得他们的信息“前后矛盾”，“有时还很幼稚”。她还说，奥特曼计划从OpenAI休假10天，以思考他对布罗克曼和苏茨克维的信任程度，以及他有多想与他们共事。
  她还说，奥特曼提到，霍顿·卡诺夫斯基——一位杰出的科技高管和有效利他主义的领导者，现在在Anthropic工作，并与Anthropic联合创始人丹妮拉·阿莫迪结婚——对“转向营利模式感到恼火，并可能提出如果OpenAI保持非营利，就提供更大笔的资金。”
  齐里斯还说，奥特曼“对于保留非营利组织做得很好”，尽管布罗克曼和苏茨克维也愿意继续采用非营利结构，“但他们知道需要提供保证，即他们不会为了让它运作而跑去搞别的事情。”

• 2017年10月，马斯克发给其Neuralink联合创始人本·拉波波特的邮件。 马斯克写道：“独立招聘或直接从OpenAI招聘。如果你向OpenAI的人推销，让他们来Neuralink工作，我没意见。”

• 2018年元旦，苏茨克维给马斯克写了一封感谢信，抄送布罗克曼。 他称马斯克是“世界上最有能力的人”，并补充说，他感谢马斯克推动OpenAI构建定制硬件。

• 2018年元旦，布罗克曼给马斯克发了与苏茨克维类似的信息。 他写道：“与您并肩工作是我的荣幸。”

• 2018年1月，马斯克、奥特曼、布罗克曼和苏茨克维之间的电子邮件往来，抄送齐里斯。 马斯克写了他对谷歌Deepmind在人工智能方面进展的担忧。他写道：“相对于谷歌，OpenAI正走在注定失败的道路上。显然需要立即采取重大行动，否则除了谷歌之外的所有人都将变得无足轻重。我考虑过首次代币发行的方式，但不会支持。在我看来，这只会导致OpenAI以及与首次代币发行相关的每个人信誉严重受损。如果某件事听起来好得不像真的，那它就不是真的。在我看来，这是一个不明智的岔路。”
  马斯克继续说：“我能想到的唯一途径是大幅扩张OpenAI和大幅扩张特斯拉的人工智能。也许两者同时进行。前者需要大幅增加捐赠资金，并让高度可信的人加入我们的董事会。目前董事会的状况非常薄弱……明确地说，我非常尊重你们的能力和成就，但我对事情的管理方式不满意。这就是我最近几个月难以与OpenAI保持接触的原因。要么我们解决问题，我大幅增加参与，要么我们不解决，我将把参与度降到几乎为零，并公开减少我的关联。我不会陷入我的影响力和时间给别人的观感与实际情况不符的境地。”
  当马斯克将这段往来转发给安德烈·卡帕斯时，卡帕斯的回复支持了马斯克的想法，他写道：“在人工智能前沿领域工作不幸非常昂贵……在我看来，今天的OpenAI正在烧钱，而且其融资模式无法达到与谷歌（一家8000亿美元的公司）认真竞争的规模。如果你不能认真竞争，却继续公开进行研究，你实际上可能是在让情况变得更糟，并且‘免费’帮助他们，因为任何进展他们都很容易复制并立即大规模整合。”
  卡帕斯继续说：“转向营利模式可能会随着时间的推移创造更可持续的收入流，并且凭借当前的团队，可能会吸引大量投资。然而，从头开始构建产品会分散人工智能研究的精力，需要很长时间，而且不清楚一家公司能否‘赶上’谷歌的规模，同时投资者可能会施加太多错误的压力。”
  卡帕斯说，他能想到的“最有希望的选择”“是让OpenAI依附于特斯拉，将其作为现金流来源。我相信，依附于其他大型企业（例如苹果、亚马逊）会因公司DNA不兼容而失败。”
  然后他详细描述了特斯拉与OpenAI合并后的样子。“用火箭来类比，特斯拉已经建造了火箭的‘第一级’，包括Model 3的整个供应链、车载计算机和持续的互联网连接。‘第二级’将是一个基于大规模神经网络训练的完全自动驾驶解决方案，OpenAI的专业知识可以极大地加速这一进程。如果在大约2-3年内实现功能完善的完全自动驾驶解决方案，我们可以销售大量的汽车/卡车。如果我们做得非常好，运输行业足够大，我们可以将特斯拉的市值提高到大约1000亿美元，并利用这些收入以适当的规模资助人工智能工作。我看不到任何其他东西有潜力在十年内达到可持续的谷歌规模资本。”
  马斯克将这条消息转发给苏茨克维和布罗克曼，写道卡帕斯是对的，并且“特斯拉是唯一有可能与谷歌一较高下的途径。即便如此，成为谷歌制衡力量的概率也很小。只是并非为零。”

• 2018年2月，马斯克与齐里斯之间的短信对话。 可能是在马斯克通过视频会议告诉奥特曼、布罗克曼和苏茨克维他将离开OpenAI董事会之后。
  齐里斯写道：“您希望我与OpenAI保持密切友好的关系以保持信息流通，还是开始疏远？信任游戏即将变得棘手，因此任何关于如何正确行事的指导都深表感谢。”马斯克回应：“保持密切友好，但我们会积极尝试从OpenAI挖三到四个人到特斯拉。随着时间的推移，会有更多人加入，但我们不会主动招募他们。”
  两人讨论了团队中可能招募的人选，齐里斯说，在马斯克离开视频会议后，苏茨克维“明显很沮丧”，并且“如果你想要伊利亚，有一定概率能得到他，但不知道你是否想要。他一直是一个非常好的精神领袖。”马斯克回应：“如果我专注于特斯拉的人工智能，OpenAI成为一股严肃力量的机会很小。”
  齐里斯接着谈到了经常被提起的对谷歌在人工智能竞赛中进展的担忧，并试图鼓励马斯克“拖慢”谷歌Deepmind的CEO哈萨比斯。她写道：“如果没有人拖慢德米斯，美好未来的概率非常低。拖慢他是我能看到唯一的、不可协商的纯粹善举。你没有意识到你有多大的能力可以直接影响他，或以其他方式拖慢他。我想你知道我不是一个恶毒的人，但在这个案例中，如果不设法拖慢或改变他的道路，感觉从根本上是不负责任的。”马斯克回应：“我怀疑我能否以有意义的方式做到这一点，”并说他们可以在当晚晚些时候通电话讨论。

• 2018年4月，马斯克与齐里斯之间的电子邮件往来。 齐里斯写道，OpenAI的首轮融资“将主要是里德·霍夫曼（LinkedIn联合创始人）的钱，可能还有一些公司。”齐里斯还写道，Quora CEO亚当·德安吉洛准备接替马斯克在OpenAI董事会的位置。（德安吉洛后来卷入了2023年奥特曼被解除CEO职务的事件。）

• 2018年7月，齐里斯给马斯克的邮件。 她向马斯克通报了OpenAI正在筹划的新一轮融资，以及未来生命研究所计划很快发布的一封关于自主武器担忧的公开信，马斯克过去曾被列为该信的签署人。
  齐里斯还转述了她听到的关于谷歌Deepmind哈萨比斯的传闻，写道：“有传闻说，除了那些因为不信任德米斯会监视他们的电子邮件和谷歌聊天而秘密在推特私信里交谈的人之外，内部核心圈子的一部分人还会在伦敦一家不带手机进场的咖啡馆里开会，以避开他进行面对面讨论。这是我从奥特曼和另一位朋友那里听说的。”

• 2018年8月，奥特曼发给马斯克的邮件。 其中包含OpenAI的正式条款清单。奥特曼写道，他“目前的想法”是不在OpenAI持有任何股权。他接着说：“我做这件事反正也不是为了钱，我喜欢完全无利益冲突、只专注于为世界带来最佳结果的想法。如果在某个时候，我们看起来不会构建通用人工智能，而是会构建一些有价值的东西，那么也许那时我会想要股权。”
  条款清单顶部有一个大的紫色警告框，里面用星号标出：“投资OpenAI有限合伙企业是一项高风险投资。投资者可能会损失其资本出资，且看不到任何回报。明智的做法是以捐赠的精神来看待对OpenAI有限合伙企业的任何投资，并理解在后通用人工智能世界中，金钱将扮演什么角色可能很难说清。”条款清单接着总结了计划的收入以及技术未来可能如何商业化，以及公司的信托责任和计划的筹资活动。
  “我们对这些原则的职责以及对推进我们使命的承诺，优先于任何产生利润的义务，”条款清单指出。“我们可能永远不会盈利，并且我们没有义务这样做。我们可以自由地将全部或部分现金流再投资于研发活动和/或相关费用，而对有限合伙人没有任何义务……非营利组织董事会的信托责任仅流向非营利组织，而非有限合伙人。”

• 2018年11月，马斯克在给视频游戏开发商Valve联合创始人加布·纽维尔的邮件中写道，他目前在OpenAI的参与度“非常有限”。
  “我仍然提供一些财务支持，并每隔几周从萨姆·奥特曼那里得到口头和邮件更新，但我不在那里花时间，”马斯克说。“我对OpenAI能否聚集起资源来有效制衡谷歌/Deepmind失去了信心，并决定改为通过特斯拉来尝试。我们有数十亿美元级别的现金流来构建硬件，希望至少能有个爆冷门的机会让谷歌保持诚实。也许值得找个时间谈谈。”
  纽维尔回应说，他很乐意在马斯克准备好时谈论特斯拉和人工智能。

• 2018年12月，马斯克与奥特曼之间的电子邮件往来，抄送了其他人。 马斯克写了他对谷歌Deepmind的哈萨比斯在人工智能竞赛中占主导地位的日益加剧的担忧。“我对OpenAI在没有任何执行力和资源上的重大改变下，能与DeepMind/谷歌相提并论的概率评估是0%。不是1%。我希望不是这样。即使筹集数亿美元也不够。这需要立即每年数十亿美元，否则就拉倒吧。不幸的是，人类的未来掌握在德米斯手中……而且他们所做的远不止这些。”
  马斯克继续说：“OpenAI让我想起了贝索斯和蓝色起源。他们绝望地落后于SpaceX，而且情况越来越糟，但贝索斯的自负让他疯狂地以为他们没有落后！我真的希望我是错的。”
  奥特曼回复询问两人能否会面讨论如何提高这个百分比。他说他相信OpenAI有一个好计划和一个好途径来获得所需的资金，但他们执行得不够快。“我们这里没有人想成为贝索斯！”他说。
  马斯克写道：“OpenAI不是DeepMind/谷歌的严肃制衡力量，而且只会进一步落后。令人惊讶的是，这……对你来说并不明显。总的来说，总是高估竞争对手。你正在做相反的事。”
  两人同意当周晚些时候在波多黎各会面。

• 2019年3月，奥特曼与马斯克之间的电子邮件往来，抄送齐里斯和特勒。 奥特曼发送了一篇详细介绍OpenAI新的利润上限结构的博客文章给马斯克批准。

• 2019年3月，齐里斯对奥特曼上述备忘录的回复。 她强调了其中提到马斯克于2018年2月离开OpenAI非营利组织董事会、且不参与OpenAI有限合伙企业事务的部分。

• 2019年3月下旬，奥特曼给马斯克发短信。 提醒他他们计划明天宣布OpenAI的新结构，并想核对一下关于马斯克过去参与度的措辞。
  “还有一些关于德米斯的不太重要的更新要分享，”奥特曼写道。马斯克同意很快通电话。

• 2019年4月，奥特曼给马斯克发短信。 询问他是否有时间谈论微软对OpenAI的投资。

• 2020年9月，马斯克公开回应一条指向VentureBeat文章的社交媒体帖子。 该文章称微软获得了OpenAI的GPT-3的独家许可。马斯克写道：“这看起来确实与‘开放’相反。OpenAI基本上被微软掌控了。”

• 2020年10月，马斯克与奥特曼之间的测试短信往来。 奥特曼主动联系说，他看到了马斯克上周在社交媒体上关于微软获得OpenAI GPT-3独家许可的帖子。奥特曼写道：“我认为，如果没有数十亿美元，我们根本无法与DeepMind相提并论，而微软似乎仍是我们在最小妥协下获得这笔资金的最佳途径。我们给了微软一份GPT-3的副本用于他们自己的产品，但我们仍然保留自主发布我们自己工作的自主权（例如，我们可以并且将继续为所有人提供访问现存最强大语言模型的API）。”
  马斯克回应：“是啊，我们该谈谈。我认为成为一个（或至少看起来是）虚伪的人不是一个成功的做法。至少改个名字。”
  马斯克后来链接到一篇社交媒体帖子，称马斯克“最严重的失误之一”就是将GPT-3独家授权给微软。奥特曼回应说，OpenAI“终于有了一个全职公关人员”，并提到了苹果前公关人员史蒂夫·道林作为新员工的名字，写道：“我希望我们能把公关做好……”道林后来在2021年初辞去了这个直接向奥特曼汇报的职务。

• 2020年10月下旬，马斯克与奥特曼之间的短信往来。 奥特曼就OpenAI正在考虑的下一轮微软投资征求意见。马斯克回应说他可以在接下来的一两天内通话。

• 2022年10月，The Information的一篇关于OpenAI与微软就额外融资进行深入谈判的文章。

• 2022年10月，马斯克在给奥特曼的短信中写道，他“很不安地看到OpenAI估值达到200亿美元……我提供了几乎所有的种子轮、A轮和大部分B轮融资。” 他发送了上述文章的链接，并补充说：“这是一种诱骗销售策略。”
  奥特曼回应：“我同意这感觉很糟糕——当我们设立利润上限结构时，我们向你提供了股权，你当时不想要，但我们仍然非常乐意在你想要的时候随时提供。鉴于我们所需的资金量以及我们仍然需要保留‘将通用人工智能交给人类’的途径，除了利润上限结构，我们看不到其他选择。顺便说一句，我个人没有股权，也从来没有过。我正在尽我所能走好这条岌岌可危的钢丝。”两人同意在接下来的一周内找个时间谈谈。

• 2023年3月，马斯克在社交媒体上发帖：“我仍然困惑，一个我捐赠了约1亿美元的非营利组织，是如何变成一个市值300亿美元的营利组织的。如果这是合法的，为什么不是每个人都这么做？”

• 2023年5月，马斯克、奥特曼、伯查尔和马斯克律师亚历克斯·斯皮罗之间的短信往来。 其中详细说明，斯皮罗，可能还有伯查尔，将前往OpenAI总部审查关于OpenAI结构及其与微软关系的文件。
  马斯克写道：“关键是理解所有公司与原始OpenAI 501c3之间的关系……了解微软拥有什么权利很重要。我担心的一件事是，他们将事实控制通用人工智能。”

• 2026年3月，马斯克的一条社交媒体帖子。 他写道：“特斯拉将成为制造通用人工智能的公司之一，并且可能是第一个以人形/原子塑造形式实现它的公司。”

• 证据编号1293
  马斯克诉奥特曼等人案中的“无争议事实”清单，包括时间线以及筹集和/或捐赠金额的细节。

2026年5月1日公布的文件

• 一份建立名为“Musk Charitable at Vanguard Charitable”的慈善账户的协议，由埃隆·马斯克于2014年7月签署。

• 萨姆·奥特曼发给埃隆·马斯克的一封电子邮件，其中包含一份OpenAI的建议清单，包括一个由五人组成的治理结构，成员包括马斯克、奥特曼、比尔·盖茨、皮埃尔·奥米迪亚和达斯汀·莫斯科维茨。“全部同意，”马斯克回复。

• 2015年12月11日题为“Introducing OpenAI”的博客文章——也可在网上公开获取。该文章将OpenAI描述为一家“非营利人工智能研究公司”，其目标是“以最有可能造福全人类的方式实现高级数字智能，不受产生财务回报需求的约束。”它列出了创始团队，包括苏茨克维、布罗克曼和安德烈·卡帕斯（后来去了特斯拉），以及联合主席奥特曼和马斯克。

• 2016年1月的一封邮件链。马斯克转发给苏茨克维和奥特曼一封来自谷歌哈萨比斯的消息，哈萨比斯反对马斯克、奥特曼和其他人“宣扬开源人工智能的优点……我猜你意识到这并非某种能神奇解决人工智能问题的万能药？”哈萨比斯称这种方法“实际上非常危险”，并链接到一篇Slate Star Codex博客文章。
  苏茨克维回应说：“随着我们越来越接近构建人工智能，开始减少开放是有意义的”，并且“不分享科学是完全没问题的（尽管为了招聘目的，短期内甚至中期内，分享一切无疑是对的）。”马斯克回复：“是的。”

• 一系列涉及马斯克、奥特曼、前OpenAI首席运营官克里斯·克拉克和罗纳德·龚（马斯克的一名助理，名字出现在财务文件上）的电子邮件。该邮件链始于2016年2月，奥特曼给马斯克发邮件说：“我认为我们需要比我最初预算更多的资金，因为a）这个领域的薪水，以及b）你想要发展的速度。”马斯克同意在未来三年内每年捐款2000万美元，而奥特曼每年捐款1000万美元，其他捐助者每年捐款500万美元。龚和克拉克讨论使用YC Org作为OpenAI的财务赞助商，克拉克附上了该组织的公司章程和其他文件。

• 一系列从Musk Charitable Fund向OpenAI提供资助的文件，包括2016年中向YC Org提供的500万美元资助，用于“OpenAI人工智能研究项目”；2016年8月的一笔450万美元资助；以及2017年一系列每月17.5万美元的租金支付等。

• 一封来自克里斯·克拉克（列为YC Org的财务官）的信函，确认马斯克于2016年5月捐赠了50万美元。

• 2016年5月布罗克曼与马斯克之间的交流。布罗克曼写道：“谷歌的政策人员想和我谈谈”，显然是因为他们担心“会构建一种认为任何闭源人工智能都是错误的公众叙事。”布罗克曼说他计划告诉他们没有必要那样做。“我们不反对人们保留专利——用它赚钱没问题，甚至有一天我们自己也可能产生收入，”他说。“什么，这很有趣。谁从谷歌打来的？”马斯克问。

• 2016年6月埃隆·马斯克与其解决问题专家贾里德·伯查尔之间的交流，讨论旧金山先锋大厦的租约（OpenAI在此办公直至2024年，之后xAI入驻）。伯查尔提到租约已敲定，正等待萨姆·奥特曼签署，马斯克表示反对：“既然是我个人承担责任，这应该被视为马斯克基金会的一栋大楼，我们将在此安置OpenAI、Neuralink，可能还有一些SpaceX或特斯拉的人。我不想让萨姆出现在租约上。”伯查尔说他会指示将奥特曼的名字从租约中移除。

• 2016年6月涉及贾里德·伯查尔、Bridgeton Holdings的两位合伙人阿蒂特·贾里瓦拉和布尔克·李的邮件链。这些信息协商了先锋大厦的租赁事宜，最后是支付首笔约14.2万美元月租的指示。

• 2016年6月奥特曼、伯查尔和克拉克之间关于为先锋大厦租约提供资金的电子邮件往来。克拉克向伯查尔发送了一份由奥特曼签署的“任意租赁协议”，附在邮件中。

• 7月1日伯查尔发给马斯克的邮件，附有已执行的先锋大厦租约，包括租约本身。伯查尔指出，大楼业主将“在我们希望的时候尽快安排现场检查”。

• 2016年7月马斯克与伯查尔之间的邮件链。伯查尔向马斯克发送了OpenAI的季度捐款和月租金详情，以及克拉克的一个请求，克拉克“问我是否可以将大楼的多余空间用于一些Y Combinator公司。”马斯克的回复中提到：“我几乎没有精力去思考这家公司，有点担心它正被当作Y Combinator的一个延伸来管理，”并说他希望将大楼的一部分用于Neuralink，“所以不要给YC用。”
  伯查尔随后说第一笔季度捐款出了问题：“因为他们连一个

英文来源：

The Musk v. Altman trial is underway, and that means exhibits, or the evidence to be presented in court, are being revealed piece by piece. So far, email exchanges, photos, and corporate documents are circulating from the earliest days of OpenAI — and from before the AI lab even had a name. Some high-level takeaways: Nvidia CEO Jensen Huang gave OpenAI an in-demand supercomputer, Musk largely drafted OpenAI’s mission and heavily influenced its early structure, OpenAI CEO Sam Altman appeared to want to lean heavily on Y Combinator for early support for OpenAI, OpenAI president Greg Brockman and Ilya Sutskever worried about Musk’s level of control over the company, and Musk highlighted the importance of a nonprofit with a mission of broadly beneficial AI.
All the evidence revealed so far in Musk v. Altman
Emails going as far back as 2015 give a glimpse into the foundations of OpenAI and the early tensions at the company.
All the evidence revealed so far in Musk v. Altman
Emails going as far back as 2015 give a glimpse into the foundations of OpenAI and the early tensions at the company.
Musk’s buzzy lawsuit, which began its jury trial on Monday in a federal courtroom in California, names Altman, Brockman, and OpenAI investor Microsoft as defendants. The claims vary against each party and have included breaching OpenAI’s charitable trust, fraud, and unjust enrichment. But ultimately, Musk’s lawsuit boils down to whether or not OpenAI deviated from its founding mission of ensuring that artificial general intelligence — an often vaguely defined term that denotes AI systems that equal or surpass human intelligence — benefits all of humanity. It’s the latest in a yearslong string of legal actions against OpenAI and its executives by Musk, who cofounded the AI lab alongside Altman and Brockman and was an early investor. (Musk also owns xAI, an AI lab that directly competes with OpenAI, and is owned by parent company SpaceX.)
Former OpenAI employees and people close to both companies have been watching this particular lawsuit with a close eye, since the outcome of a jury trial could have affected how OpenAI runs its business and controls its quickly advancing technology. Plus, OpenAI and SpaceX are both reportedly racing to go public this year, so they’re more in the public eye than ever.
The lawsuit discovery process had already unearthed a lot of eyebrow-raising communications between AI industry executives, from emails between Altman and Sutskever to entries from Brockman’s own diary. Even texts between Meta CEO Mark Zuckerberg and Musk were made public. But that was all before the jury trial started — now, there’s even more set to be revealed.
Here’s an exhaustive list of all the exhibits that have been made public so far and the biggest takeaways from each one. Admittedly, not every item is necessarily interesting, so we’ve flagged the most important ones with an asterisk. The Verge will keep updating the list as more are added.
Documents released April 29, 2026
A June 2015 email exchange between Altman and Musk. Altman lays out a five-part plan involving an AI lab with a mission to “create the first general AI and use it for individual empowerment—ie, the distributed version of the future that seems the safest. More generally, safety should be a first-class requirement.”
He suggests that they start with seven to 10 people and expand from there, using an extra Y Combinator building located in Mountain View. Governance-wise, Altman names five people to start, proposing himself, Musk, Bill Gates, Pierre Omidyar, and Dustin Moskovitz. “The technology would be owned by the foundation and used ‘for the good of the world’, and in cases where it’s not obvious how that should be applied the 5 of us would decide,” Altman writes. He adds that the researchers working at the lab would have “significant financial upside … uncorrelated to what they build, which should eliminate some of the conflict,” and suggests paying them a “competitive salary” and awarding them equity in Y Combinator. He also says they should get someone to “run the team” but that that person “probably shouldn’t be on the governance board.”
Altman goes on to ask Musk whether he’ll be involved in the AI lab in addition to governance, potentially coming by once a month to talk about progress or at least being publicly supportive to help with recruiting. As a model, he names Peter Thiel’s “part-time partner” involvement at Y Combinator.
Finally, Altman mentions a “regulation letter,” seeming to imply that the AI lab was going to write a letter calling for AI regulation. He says he’s happy to leave Musk off as a public signatory.
Musk replies, “Agree on all.”
In an October 2015 email exchange between Altman and Musk, Altman suggests starting with a $100 million commitment by Musk and asks if he could donate an additional $30 million over the next five years. He says Bill Gates isn’t yet committed to donating but that he hopes to “have him locked down next week,” adding that he believes Mark Zuckerberg likely won’t come through due to his own AI lab, Facebook AI Research (FAIR). He also suggests that he and Musk start as the first two members of the Safety Board with the potential to add three other members over the following year, calling it the “‘second key’ for releasing anything that could be dangerous.”
Musk responds, “Let’s discuss governance. This is critical. I don’t want to fund something that goes in what turns out to be the wrong direction.”
In a November 2015 email exchange between Musk and Altman, the two discuss plans for the forthcoming AI lab. Musk starts off by recounting a “great call with Greg [Brockman]” and saying he’s “super impressed with everyone so far,” calling it a “great team.” He suggests creating the lab as an “independent, pure play 501c3, but with a crystal clear focus on the positive advent of strong AI distributed widely to humanity.” He says the company would “still aim to bring in revenue in excess of costs at some point, but positive net revenue would just flow to cash reserves.”
With regard to compensation for employees, Musk suggests a cash salary and certain bonuses. He says that if Altman is amenable, employees could convert cash to stock in Y Combinator, adding that it’s fine if they’d rather convert some or all to SpaceX stock instead. (“I can pretty much do what I want on the SpaceX side, as it is private (thank goodness),” Musk writes.) He also offers “insane amounts of real world sensor data” from Tesla for the AI lab to use, mentioning that the amount of data is “several orders of magnitude greater than any other company.”
Musk’s first stab at a name for the AI lab is “Freemind,” as he says it “conveys the sense that we are trying to create digital intelligence that will be freely available to all — the opposite of Deepmind’s one-ring-to-rule-them-all approach.” He also says he’ll dedicate whatever amount of his time is useful, even though that could mean less time allocated to SpaceX and Tesla. “If I really believe that this is potentially the biggest near-term existential threat, then action should follow belief,” he writes. He adds later that, despite seemingly trying to be essentially a silent partner, he has to “bite the bullet on admitting real involvement. This will come as a shocker to many, but so be it. Can’t be lukewarm about this.”
Altman suggests the AI lab share a building with Y Combinator and use the incubator’s legal team to help get it started. He also suggests the names “Axon” or something related to famed computer scientist and mathematician Alan Turing.
Musk writes, “Something Turing-related that doesn’t sound too ominous might be good. Want to avoid the Turing Test association though, as that sounds too much like we are replacing humans.”
A December 2015 email exchange between Altman and Musk drafts the opening paragraphs of OpenAI’s mission and press release. Musk says the “whole point of this release is to attract top talent.” The two go back and forth on wording, and the final product ends up not straying too much from Musk’s original draft.
Musk writes in his draft that “the outcome of this venture is uncertain and the pay is low compared to what others will offer, but we believe the goal and the structure are right.” Altman writes in his draft that “because we don’t have any financial obligations, we can focus on the maximal positive human impact and disseminating AI technology as broadly as possible.”
OpenAI’s official articles of incorporation, filed December 8th, 2015. The document states that OpenAI “shall be a nonprofit corporation organized exclusively for charitable purposes” and that its purpose is “to ensure that artificial general intelligence benefits all of humanity, including by conducting and/or funding artificial intelligence research. The corporation may also research and/or otherwise support efforts to safely develop and distribute such technology and its associated benefits, including analyzing the societal impacts of the technology and supporting related educational, economic, and safety policy research and initiatives.”
The document continues, “The resulting technology will benefit the public and the corporation will seek to distribute it for the public benefit when applicable. The corporation is not organized for the private gain of any person.”
An April 2016 email exchange between Musk and Nvidia CEO Jensen Huang. Musk asks Huang if the OpenAI team can buy an early unit of a supercomputer, making sure to highlight that “OpenAI is unaffiliated with Tesla. It is a non-profit funded by me and a few others with the goal of developing safe AGI (and hopefully not paving the road to hell with good intentions).”
Huang responds that he “will make sure OpenAI gets one of the first ones.”
A photo of Jensen Huang ostensibly dropping off said computer. Elon Musk stands nearby.
On the wall behind him is a lengthy quote sometimes attributed to US Navy Admiral Hyman G. Rickover, which is echoed in a 2013 blog post by Altman. (The Verge couldn’t immediately confirm the whole quote was said by Rickover; in a US Navy post attributed to the admiral, only part of the quote appears: “Man has a large capacity for effort. But it is so much greater than we think it is, that few ever reach this capacity.”)
In an August 2017 email exchange between Musk and Shivon Zilis, Musk’s chief of staff who eventually sat on OpenAI’s board, and with whom Musk would eventually share multiple children. Zilis writes a recap of her meeting with Brockman and Sutskever, laying out seven unanswered questions. She says Brockman and Sutskever are fine with Musk spending less time on the company and having less control, or spending more time and having more control, but not less time and more control. They also hope to raise significantly more than $100 million to start, as they worry the data center they need alone would cost that much. She says Brockman is relatively set on an equal equity split. They also, she writes, worry about Musk’s control over the company. In her notes recapping their concerns, Zilis writes, “Is the requirement for absolute control? They wonder if there is a scenario where there could be some sort of creative overrule position if literally everyone else disagreed on direction.”
The biggest point of tension, Zilis writes, seems to be on Musk’s duration of control over the company, despite his ownership stake. “The non-negotiable seems to be an ironclad agreement to not have any one person have absolute control of AGI if it’s created. Satisfying this means a situation where, regardless of what happens to the three of them [Greg, Ilya, and Sam], it’s guaranteed that power over the company is distributed after the 2-3 year initial period … An ironclad 2-3yr minority control agreement, regardless of the fates of Greg / Sam / Ilya.”
Musk responds, “This is very annoying. Please encourage them to go start a company. I’ve had enough.”
A September 2017 email to Musk from Jared Birchall, an adviser to Musk and manager of his family office. He attaches a “more user friendly version of the cap table that Ilya and Greg are proposing.”
In it, Musk is reflected as having 51.20 percent equity, with Altman, Sutskever, and Brockman each having 11.01 percent. There’s also reserved equity for employees, and the cap table denotes each initial employee’s name or nickname followed by a proposed amount of equity.
Documents released April 30, 2026
A November 2015 email exchange between Musk and Altman, in which Altman references what seems to be one of the first names and structures considered for the AI lab — Y Combinator AI.
Altman writes that the “plan is to have you, me, and Ilya on the Board of Directors for YC AI, which will be a Delaware non-profit,” adding, “We will write into the bylaws that any technology that potentially compromises the safety of humanity has to get consent of the Board to be released, and we will reference this in the researchers’ employment contracts.”
Musk disagrees in his response: “I think this should be independent from (but supported by) YC, not what sounds like a subsidiary. Also, the structure doesn’t seem optimal. In particular, the YC stock along with a salary from the nonprofit muddies the alignment of incentives. Probably better to have a standard C corp with a parallel nonprofit.”
In a December 2016 email exchange between Musk and his Neuralink associates, he brings up his concerns about beating Google Deepmind again, writing, “Deepmind is moving very fast. I am concerned that OpenAI is not on a path to catch up. Setting it up as a non-profit might, in hindsight, have been the wrong move. Sense of urgency is not as high.”
In June 2017, Musk writes an email saying he hired Andrej Karpathy away from OpenAI to be director of Tesla Vision, saying, “The OpenAI guys are gonna want to kill me, but it had to be done…”
In July 2017, Musk writes in an email to Sutskever and Brockman that China “will do whatever it takes to obtain what we develop. Maybe another reason to change course.” Brockman says he agrees, and that the path ahead should be an “AI research non-profit (through end of 2017), AI research and hardware for-profit (starting 2018), [and] government project (when: ??).”
As a token of appreciation for their work at OpenAI, Musk offers to give Sutskever, Brockman, and others on the team Tesla Model 3 cars that are “not available to the public.”
Musk asks in August 2017 if Altman, Sutskever, and Brockman can meet to discuss the “next step” for OpenAI — and volunteers “the haunted mansion [he] just bought near SF,” although it’s “kinda crazy and weird and will have party carnage.”
An email exchange between Musk and Birchall, his money manager, later in August 2017. Birchall writes that for now, he’s “held off” on giving OpenAI Musk’s typical quarterly $5 million donation and asks if he should continue holding off. Musk responds affirmatively.
A September 2017 email exchange between Musk, Brockman, and Sutskever, with Sutskever suggesting that Musk have three board seats and Brockman, Sutskever, and Altman each have one. Musk responds that he believes he should have the right to appoint four board seats and later compliments the three others.
Musk writes, “I would not expect to appoint [the four board seats] immediately, but, like I said I would unequivocally have initial control of the company, but this will change quickly. The rough target would be to get to a 12 person board (probably more like 16 if this board really ends up deciding the fate of the world) where each board member has a deep understanding of technology, at least a basic understanding of AI and strong & sensible morals.”
A September 2017 email exchange between Brockman and Musk, with Altman and Sutskever CC’d. Brockman and Sutskever propose a cap table for Musk’s approval, with Brockman noting that himself and Altman are able to invest a lot more than Sutskever, but Sutskever can invest more than $2.5 million if he takes a loan from Altman and/or Brockman securitized by stock he owns.
Musk replies, “Guys, you are pushing too hard here. I’m not ok with this.”
A September 2017 text message from Musk to Zilis and others. Musk writes, “We should get going on creating the OpenAI B Corp, as I promised Greg and Ilya. Let’s discuss this eve. Still no word from Sam Altman btw.”
A September 2017 email exchange between Altman, Musk, Zilis, Brockman, Sutskever, and Musk’s chief of staff Sam Teller. It paints a picture of a two-sided negotiation with peak tension, with Musk and Altman essentially on one side and Brockman and Sutskever on the other.
To Elon, Brockman and Sutskever write, “Elon: We really want to work with you … Our desire to work with you is so great that we are happy to give up on the equity, personal control, make ourselves easily firable — whatever it takes to work with you.” However, they write they were concerned about Musk’s control over the future technology OpenAI may put out.
“The current structure provides you with a path where you end up with unilateral absolute control over the AGI,” the two write to Musk. “You stated that you don’t want to control the final AGI, but during this negotiation, you’ve shown to us that absolute control is extremely important to you. As an example, you said that you needed to be CEO of the new company so that everyone will know that you are the one who is in charge, even though you also stated that you hate being CEO and would much rather not be CEO. Thus, we are concerned that as the company makes genuine progress towards AGI, you will choose to retain your absolute control of the company despite current intent to the contrary. We disagree with your statement that our ability to leave is our greatest power, because once the company is actually on track to AGI, the company will be much more important than any individual.”
The two also touch on the team’s often-mentioned fears about Deepmind’s Demis Hassabis. To Musk, they write, “The goal of OpenAl is to make the future good and to avoid an AGI dictatorship. You are concerned that Demis could create an AGI dictatorship. So do we. So it is a bad idea to create a structure where you could become a dictator if you chose to, especially given that we can create some other structure that avoids this possibility.”
Brockman and Sutskever have different concerns for Altman himself, though.
In the part of the message directed at Altman, they write, “We haven’t been able to fully trust your judgements throughout this process, because we don’t understand your cost function. We don’t understand why the CEO title is so important to you. Your stated reasons have changed, and it’s hard to really understand what’s driving it.” Separately, they question some of Altman’s motivations, asking him, “Is AGI truly your primary motivation? How does it connect to your political goals? How has your thought process changed over time?”
Altman responded to the email that he “remain[ed] enthusiastic about the non-profit structure!”
A September 2017 response from Musk to the above concerns detailed by Brockman and Sutskever. Musk writes, “Guys, I’ve had enough. This is the final straw. Either go do something on your own or continue with OpenAl as a nonprofit. I will no longer fund OpenAl until you have made a firm commitment to stay or I’m just being a fool who is essentially providing free funding for you to create a startup. Discussions are over.”
A September 2017 email exchange between Zilis and Musk. Zilis recounts some of Altman’s feelings, like the idea that Altman “lost a lot of trust” for Brockman and Sutskever during the negotiations, feeling that their messaging was “inconsistent” and “childish at times.” She also says Altman was planning to take a 10-day hiatus from OpenAI to think about how much he trusted Brockman and Sutskever and how much he wanted to work with them.
She also says Altman mentioned that Holden Karnofsky — a prominent tech executive and leader in effective altruism, who now works at Anthropic and is married to Anthropic co-founder Daniela Amodei — was “irked by the move to for-profit and potentially offered [a] more substantial amount of money if OpenAI stayed a non-profit.”
Zilis also says that Altman is “great with keeping the non-profit” and that though Brockman and Sutskever are also amenable to continuing with the non-profit structure, “they know they would need to provide a guarantee that they won’t go off doing something else to make it work.”
An October 2017 email from Musk to his Neuralink co-founder Ben Rapoport. Musk writes, “Hire independently or directly from OpenAI. I have no problem if you pitch people at Open Al to work at Neuralink.”
On New Year’s Day in 2018, Sutskever writes a note of gratitude to Musk, cc’ing Brockman, calling Musk the “most overwhelmingly competent person in the world” and adding that he’s thankful Musk pushed OpenAI to build custom hardware.
Brockman sends a similar message as Sutskever did to Musk on New Year’s Day 2018, writing that “it’s an honor to work alongside you.”
In a January 2018 email exchange between Musk, Altman, Brockman, and Sutskever, with Zilis CC’ed, Musk writes of his concerns about Google Deepmind’s advancement in AI. He writes, “OpenAl is on a path of certain failure relative to Google. There obviously needs to be immediate and dramatic action or everyone except for Google will be consigned to irrelevance. I have considered the ICO approach and will not support it. In my opinion, that would simply result in a massive loss of credibility for OpenAl and everyone associated with the ICO. If something seems too good to be true, it is. This was, in my opinion, an unwise diversion.”
Musk continues, “The only paths I can think of are a major expansion of OpenAl and a major expansion of Tesla Al. Perhaps both simultaneously. The former would require a major increase in funds donated and highly credible people joining our board. The current board situation is very weak … To be clear, I have a lot of respect for your abilities and accomplishments, but I am not happy with how things have been managed. That is why I have had trouble engaging with OpenAl in recent months. Either we fix things and my engagement increases a lot or we don’t and I will drop to near zero and publicly reduce my association. I will not be in a situation where the perception of my influence and time doesn’t match the reality.”
When Musk forwards the back-and-forth to Andrej Karpathy, Karpathy responds in support of Musk’s thoughts, writing, “Working at the cutting edge of AI is unfortunately expensive … It seems to me that OpenAl today is burning cash and that the funding model cannot reach the scale to seriously compete with Google (an 800B company). If you can’t seriously compete but continue to do research in open, you might in fact be making things worse and helping them out ‘for free,’ because any advances are fairly easy for them to copy and immediately incorporate, at scale.”
Karpathy continues, “A for-profit pivot might create a more sustainable revenue stream over time and would, with the current team, likely bring in a lot of investment. However, building out a product from scratch would steal focus from Al research, it would take a long time and it’s unclear if a company could ‘catch up’ to Google scale, and the investors might exert too much pressure in the wrong directions.”
Karpathy says the “most promising option” he can think of “would be for OpenAl to attach to Tesla as its cash cow. I believe attachments to other large suspects (e.g. Apple? Amazon?) would fail due to an incompatible company DNA.”
He then goes on to detail what a Tesla-OpenAI merge would look like. “Using a rocket analogy, Tesla already built the ‘first stage’ of the rocket with the whole supply chain of Model 3 and its onboard computer and a persistent internet connection. The ‘second stage’ would be a full self driving solution based on large-scale neural network training, which OpenAl expertise could significantly help accelerate. With a functioning full self-driving solution in ~2-3 years we could sell a lot of cars/trucks. If we do this really well, the transportation industry is large enough that we could increase Tesla’s market cap to high O(~100K), and use that revenue to fund the Al work at the appropriate scale. I cannot see anything else that has the potential to reach sustainable Google-scale capital within a decade.”
Musk forwards the note to Sutskever and Brockman, writing that Karpathy is right, and that “Tesla is the only path that could even hope to hold a candle to Google. Even then, the probability of being a counterweight to Google is small. It just isn’t zero.”
A February 2018 text message conversation between Musk and Zilis, potentially just after Musk told Altman, Brockman, and Sutskever on a video meeting that he would be departing OpenAI’s board.
Zilis writes, “Do you prefer I stay close and friendly to OpenAl to keep info flowing or begin to disassociate? Trust game is about to get tricky so any guidance for how to do right by you is appreciated.” Musk responded, “Close and friendly, but we are going to actively try to move three or four people from OpenAl to Tesla. More than that will join over time, but we won’t actively recruit them.”
The two discuss who on the team to potentially recruit, with Zilis saying that Sutskever was “visibly devastated” after Musk left the video meeting and that there is “some probability you could get Ilya if you wanted him, but don’t know if you do. He has been a very good spiritual leader.” Musk responds, “There is little chance of OpenAI being a serious force if I focus on Tesla AI.”
Zilis goes on to touch on the often-brought-up fear of Google’s progress in the AI race and tries to encourage Musk to “slow down” Hassabis, CEO of Google Deepmind. She writes, “There is a very low probability of a good future if someone doesn’t slow Demis down. Slowing him down is the only nonnegotiable net good action I can see. You don’t realize how much you have an ability to influence him directly or otherwise slow him down. I think you know I’m not a malicious person but in this case it feels fundamentally irresponsible to not find a way to slow or alter his path.” Musk responds, “I doubt I could do so in a meaningful way,” and says they can speak by phone about it later that evening.
An April 2018 email exchange between Musk and Zilis, with Zilis writing that OpenAI’s first funding round will likely be “largely Reid [Hoffman, LinkedIn co-founder] money, potentially some corporates.” Zilis also writes that Quora CEO Adam D’Angelo is primed to take Musk’s place on OpenAI’s board. (D’Angelo would later be involved in Altman’s 2023 ouster from his CEO role.)
In a July 2018 email to Musk, Zilis updates him on the new funding round OpenAI is planning, as well as a public letter detailing concerns about autonomous weapons that the Future of Life Institute is planning to publish soon, which Musk had been listed as a signatory on in the past.
Zilis also recounts rumors she’s heard about Google Deepmind’s Hassabis, writing, “Rumor has it that, on top of the folks that secretly converse on Twitter DM because they don’t trust Demis not to spy on their email and gchat, a part of the inner group also meets in a London coffee shop without cell phones to have in person discussions away from him. Heard this from both Altman and another friend.”
An August 2018 email from Altman to Musk, in which he includes OpenAI’s official term sheet. Altman writes that his “current thought” is that he won’t take any equity in OpenAI. He goes on to say, “I’m not doing this for the money anyway, and I like the idea of being completely unconflicted and just focused on the best outcome for the world. If it appeared at some point we weren’t going to build AGI but were going to build something valuable, then maybe I’d want equity then.”
The term sheet includes a large purple warning box at the top, stating within asterisks, “Investing in OpenAI LP (the Partnership) is a high-risk investment. Investors could lose their capital contribution and not see any return. It would be wise to view any investment in OpenAI LP in the spirit of a donation, with the understanding that it may be difficult to know what role money will play in a post-AGI world.” The term sheet goes on to summarize planned revenue and how technology may be commercialized in the future, as well as the company’s fiduciary duties and planned fundraising.
“Our duty to these principles and the advancement of our mission takes precedence over any obligation to generate a profit,” the term sheet states. “We may never make a profit, and we are under no obligation to do so. We are free to re-invest any or all of our cash flow into research and development activities and/or related expenses without any obligation to the Limited Partners … The fiduciary duties of the Nonprofit Board of Directors flow exclusively to the Nonprofit, not to the Limited Partners.”
In November 2018, Musk writes in an email to Gabe Newell, co-founder of video game developer Valve, that his involvement in OpenAI is “very limited at this point.”
“I still provide some financial support and get verbal and email updates every few weeks from Sam Altman, but don’t spend time there,” Musk says. “I lost confidence that OpenAl could muster the resources to serve as an effective counterweight to Google/Deepmind and decided to attempt that through Tesla instead. We have cash flow on the order of billions of dollars per year to build hardware that hopefully has at least a dark horse chance to keep Google honest. Probably worth talking about at some point.”
Newell responds that he’s happy to talk about Tesla and AI when Musk is ready.
A December 2018 email exchange between Musk and Altman, with others CC’ed. Musk writes of his intensifying fears about Google Deepmind’s Hassabis taking over in the AI race. “My probability assessment of OpenAl being relevant to DeepMind/Google without a dramatic change in execution and resources is 0%. Not 1%. I wish it were otherwise. Even raising several hundred million won’t be enough. This needs billions per year immediately or forget it. Unfortunately, humanity’s future is in the hands of Demis … And they are doing a lot more than this.”
Musk continues, “OpenAl reminds me of Bezos and Blue Origin. They are hopelessly behind SpaceX and getting worse, but the ego of Bezos has him insanely thinking that they are not! I really hope I’m wrong.”
Altman responds to ask if the two can meet to discuss increasing that percentage. He says he believes OpenAI has a good plan and a good path to gain the capital they need but that they aren’t executing quickly enough. “None of us want to be Bezos here!” he says.
Musk writes, “OpenAl is not a serious counterweight to DeepMind/Google and will only get further behind. It is surprising that this … isn’t obvious to you. In general, always overestimate competitors. You are doing the opposite.”
The two agree to meet in Puerto Rico later that week.
A March 2019 email exchange between Altman and Musk, with Zilis and Teller CC’ed. Altman sends a blog post detailing OpenAI’s new capped-profit structure to Musk for approval.
Zilis circles back on Altman’s note above in March 2019, highlighting the part where it says Musk left the board of OpenAI’s nonprofit in February 2018 and that he is not involved with OpenAI LP.
Altman texts Musk a couple of days later in March 2019, reminding him they’re planning to announce OpenAI’s new structure tomorrow and wanting to check the wording about Musk’s past involvement.
“Also have some mild Demis updates to share,” Altman writes. Musk agrees to talk over the phone soon.
In April 2019, Altman texts Musk to ask if he has time to talk about Microsoft’s investment in OpenAI.
In September 2020, Musk publicly responds to a social media post linking to a VentureBeat article about Microsoft getting the exclusive license to OpenAI’s GPT-3, writing, “This does seem like the opposite of open. OpenAI is essentially captured by Microsoft.”
An October 2020 test message exchange between Musk and Altman, with Altman reaching out to say he saw Musk’s posts on social media the prior week about Microsoft’s exclusive license to OpenAI’s GPT-3. Altman writes, “I think there’s no way we can hold a candle to DeepMind without many billions of dollars, and MSFT still seems like the best way for us to get that with the least compromise. We gave MSFT a copy of GPT-3 to use in their own products, but we still get to retain autonomy to release our work ourselves (e.g., we can and will continue to provide API access to the most powerful language model in existence to everyone).”
Musk responds, “Yeah, we should talk. I don’t think it’s a winning approach to be (or at least appear to be) hypocritical. At least change the name.”
Musk later links to a social media post saying that one of Musk’s “worst management blunders” was exclusively licensing GPT-3 to Microsoft. Altman responds saying that OpenAI “finally just got a full time PR person,” name-dropping Apple’s former PR person Steve Dowling as the new hire, and writing, “I am hopeful we can start getting pr right…” Dowling would later step down from his role, which reported directly to Altman, at the beginning of 2021.
In a text message exchange between Musk and Altman in late October 2020, Altman asks for advice on the next Microsoft investment that OpenAI is considering. Musk responds that he can talk in the next day or two.
An October 2022 article from The Information about OpenAI’s advanced talks with Microsoft for additional funding.
In October 2022, Musk writes in a text message to Altman that he was “disturbed to see OpenAI with a $20B valuation … I provided almost all the seed, A and most of B round funding.” He sends a link to the above article, adding, “This is a bait and switch.”
Altman responds, “I agree this feels bad—we offered you equity when we established the cap profit, which you didn’t want at the time but we are still very happy to do any time you’d like. We saw no alternative, given the amount of capital we needed and needing still to preserve away to ‘give the AGI to humanity’, other than the capped profit structure. Fwiw I personally have no equity and never have. Am trying to navigate tricky tightrope the best I can.” The two agree to talk sometime in the coming week.
In March 2023, Musk posts on social media, “I’m still confused as to how a non-profit to which I donated ~ $100M somehow became a $30B market cap for-profit. If this is legal, why doesn’t everyone do it?”
A May 2023 text message exchange between Musk, Altman, Birchall, and Musk lawyer Alex Spiro, in which it’s detailed that Spiro, and potentially Birchall, will show up to OpenAI’s headquarters to review documents about OpenAI’s structure and its relationship with Microsoft.
Musk writes, “The point is to understand the relationship between all the companies and the original OpenAI 501c3 … Understanding what rights Microsoft has is important. One of the things I’m concerned about is that they will have de facto control over AGI.”
A March 2026 social media post by Musk. He writes, “Tesla will be one of the companies to make AGI and probably the first to make it in humanoid/atom-shaping form.”
Exhibit No. 1293
A list of “undisputed facts” in Musk v. Altman, et al., including details on timeline and amounts of money raised and/or donated.
Documents released May 1, 2026
An agreement establishing a philanthropic account called Musk Charitable at Vanguard Charitable, signed by Elon Musk in July of 2014.
An email from Sam Altman to Elon Musk with a list of suggestions for OpenAI, including a governance structure of five people, including Musk, Altman, Bill Gates, Pierre Omidyar, and Dustin Moskovitz. “Agree on all,” Musk responds.
A December 11th, 2015 blog post titled “Introducing OpenAI” — also available publicly online. The post describes OpenAI as a “non-profit artificial intelligence research company” whose goal is to reach “advanced digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by the need to generate financial return.” It lists the founding team, including Sutskever, Brockman, and Andrej Karpathy (who would later go to Tesla), as well as the co-chairs, Altman and Musk.
A January 2016 email chain. Musk forwards Sutskever and Altman a message from Google’s Hassabis, where Hassabis objects to Musk, Altman, and others “extolling the virtues of open sourcing AI … I presume you realise that this is not some sort of panacea that will somehow magically solve the AI problem?” Hassabis describes the approach as “actually very dangerous” and links to a Slate Star Codex blog post.
Sutskever responds by saying that “as we get closer to building AI, it will make sense to start being less open” and “totally OK to not share the science (even though sharing everything is definitely the right strategy in the short and possibly medium term for recruitment purposes).” Musk replies: “Yup.”
A series of emails involving Musk, Altman, former OpenAI COO Chris Clark, and Ronald Gong (an associate of Musk who’s listed on financial documents). The chain starts in February of 2016, with Altman emailing Musk that “I think we’re going to need more than I was originally budgeting given a) the salaries in the field and b) the speed at which you want to grow.” Musk agrees to contribute $20 million a year for the next three years, while Altman contributes $10 million a year, and $5 million a year comes from other donors. Gong and Clark discuss using YC Org as a fiscal sponsor for OpenAI, and Clark attaches the organization’s articles of incorporation and other documentation.
Documentation for a series of grants from the Musk Charitable Fund to OpenAI, including a mid-2016 grant of $5 million to YC Org, directed toward the “OpenAI Artificial Intelligence Research Program”; a $4.5 million grant in August of 2016; and a series of monthly $175,000 lease payments in 2017, among others.
A letter from Chris Clark (listed as the treasurer of YC Org) acknowledging a $500,000 donation from Musk in May of 2016.
A May 2016 exchange between Brockman and Musk. Brockman writes that “Google’s policy people want to speak with me,” apparently because they’re afraid they’ll “build a public narrative that it’s wrong to have any closed-source AI.” Brockman says he plans to say there’s no reason to do that. ”We don’t have a problem with people keeping things proprietary — it’s fine to make money off this stuff, and we may even generate revenue ourselves one day,” he says. “What, that’s really interesting. Who called from Google?” Musk asks.
A June 2016 exchange between Elon Musk and his fixer, Jared Birchall, discussing a lease of the Pioneer Building in San Francisco (which housed OpenAI until 2024 and xAI after that). Birchall mentions that a lease has been finalized and is awaiting Sam Altman’s signature, and Musk objects: “Since I’m personally on the hook, this should be viewed as a Musk Foundation building, in which we will house OpenAI, Neuralink, and maybe some SpaceX or Tesla people. I don’t want Sam on the lease.” Birchall says he’ll direct Altman’s name to be removed from the lease.
A June 2016 email chain involving Jared Birchall and two associates of Bridgeton Holdings, Atit Jariwala, and Bourke Lee. The messages negotiate leasing the Pioneer Building and end with instructions for making the first monthly lease payment of around $142,000.
A June 2016 email exchange between Altman, Birchall, and Clark about financing the Pioneer Building lease. Clark sends Birchall a Tenancy at Will agreement signed by Altman, attached to the email.
A July 1st email from Birchall to Musk with the executed lease to the Pioneer Building, including the lease. Birchall notes that the building owner will “facilitate a site inspection as soon as we’d like.”
A July 2016 email chain between Musk and Birchall. Birchall sends Musk details about the quarterly donations and monthly rent payments for OpenAI, plus a request from Clark, who “asked me about using the extra space in the building for some of the Y Combinator companies.” Musk’s response mentions that “I have had very little bandwidth to think about the company and am a little worried that it is being managed as an extension of Y Combinator” and says he’d also like to use part of the building for Neuralink, “so no YC stuff.”
Birchall then says there was a problem with the first quarterly contribution: “because they didn’t have an entity in place to even make a contribution we didn’t pay,” and in June they began using another nonprofit (presumably YC Org) as a conduit. “I’m not sure why they have taken so long to apply,” Birchall complains. “So I haven’t sent anything to OpenAI? That’s a really big deal. My credibility is at stake here,” Musk writes. Birchall confirms the funds were sent — just channeled through a temporary 501(c)(3). “Good,” Musk answers.
An August 2016 email exchange between Musk and Altman. Altman tells Musk he’s negotiated a $50 million compute donation from OpenAI over the next 3 years and asks if there’s any reason to care about switching from Amazon. “I’m ok with this only if they don’t use it in marketing. I would also like to see the exact terms and conditions. Gifts are only as good as the T&C,” Musk writes. “I think Jeff [Bezos] is a bit of a tool and Satya [Nadella] is not, so I slightly prefer Microsoft, but I hate their marketing dept.”
Altman writes that “Amazon started really dicking us around on the T+C, especially on marketing commits. And their offering wasn’t that good technically anyway.” Musk says that “I will call Satya if we get to decent terms” and says that Microsoft can always point people to “a simple text blog expressing appreciation of Microsoft’s donation on our website.”
A series of emails between October and November of 2016 involving Birchall; Gong; Morgan Stanley Private Wealth Management group director Matilda Simon-Ferrigno; and two people from Gong’s company myCFO, Teresa Holland and Paula Lo. Birchall arranges moving shares from the Musk Foundation to finance OpenAI.
OpenAI’s 2016 tax returns as a 501(c)(3) nonprofit. It lists 52 employees and around $13 million in total revenue, mostly from contributions and grants. It names accomplishments including establishing a research team, launching the OpenAI Gym Beta, publishing “nearly half a dozen comprehensive research papers,” holding a conference, and building a safety team.
The Musk Foundation’s 2016 Return of Private Foundation tax documents, showing a total of around $47.8 million in contributions, gifts, and grants.
A March 2017 letter from Chris Clark to Elon Musk, acknowledging a gift of $5 million to OpenAI via YC Org.
A June 2017 letter from Chris Clark to Elon Musk, acknowledging a gift of $5 million to OpenAI via YC Org.
A June 2017 Fidelity charitable investment advisor program application for the Musk Foundation Charitable Fund.
Emails between Birchall, Clark, and UBS wealth management associate Leeder Hsu in July of 2017. Birchall directs a grant of $250,000 to YC Org for a Universal Basic Income study.
A July 2017 email chain involving Brockman, Musk, Sutskever, and Birchall. Musk sends a link to a New York Times story about Chinese AI with the comment, “They will do whatever it takes to obtain what we develop. Maybe another reason to change course.” Brockman suggests a path of an AI research nonprofit through 2017, “AI research + hardware for-profit” starting 2018, and “Government project (when: ??).” Musk then says that “in appreciation for what you’ve done to get OpenAI to where it is today,” he’d like to offer some OpenAI founding members Tesla Founder Series Model 3 cars. Birchall says he’ll reach out with details about the cars.
An August 2017 email conversation between Zilis and Birchall about filing for a for-profit branch of OpenAI. “Elon wants to have control to prevent this from going squirrely,” Zilis says. She lists “unknowns,” including leadership of the new entity — ”Greg 100% doesn’t want to run it.” Birchall sends confirmation of how much Musk gave to OpenAI in 2016 and 2017: $15.4 million and $16 million, respectively.
OpenAI, Inc.’s certificate of incorporation on September 15th, 2017, as a public benefit corporation.
OpenAI’s 2017 tax returns, also as a 501(c)(3) nonprofit. It lists around $33 million in revenue (mostly from contributions and grants, again) and 99 employees. It notes that in 2017, it demonstrated that “reinforcement learning algorithms could be scaled to beat the world’s best humans at a restricted version of an advanced, multiplayer game called Dota 2.”
A copy of the Vanguard Charitable Policies and Guidelines, 2014 to 2017.
Documentation for a series of 2017-2020 donations from Musk to OpenAI, composed of monthly “general support” payments that likely include the Pioneer Building lease — which Musk said constituted his main form of support in the later years of OpenAI.
A January 2018 letter from Clark to Musk acknowledging a gift of four Tesla sedans with a total value of around $250,000.
The OpenAI Charter from April 9th, 2018. It outlines “the principles we use to execute on OpenAI’s mission,” including “broadly distributed benefits,” “long-term safety,” and “technical leadership.”
An August 31st, 2018 email from Altman to Musk with a for-profit Limited Partnership term sheet attached. “Please see attached, look forward to feedback,” Altman says. He says that “my current thought is that I won’t take any equity,” since he likes the idea of “being completely unconflicted,” but says that if OpenAI appeared unlikely to build AGI but “were going to build something valuable, then maybe I’d want equity then.” At the start of the term sheet is a box marked “Important warning,” saying that the partnership is a “high-risk investment” and any investment should be “in the spirit of a donation.”
An August 31st, 2018 email from Zilis to Birchall, forwarding Altman’s email. Birchall responds: “Pretty plain vanilla for-profit structure. So kinda hard to push a narrative that doesn’t involve investors being very focused on ROI. I’m a super fan of capitalism and making tons of money doing great things, but not sure if this correlates with the whole ‘noble cause for humanity, not doing it to make money’ narrative.”
A July 2020 email from Clark to Birchall confirming that OpenAI’s for-profit entity will take over rent payments and suggesting a final one-time donation for security costs and “anticipated landlord project passthrough” of $570,000. “We certainly understand if you’d prefer to just stop everything now,” Clark says, telling Birchall to “do whatever you feel is most fair.”
A November 2018 text message chain between Birchall and Greg Smithies, then Neuralink and the Boring Company’s finance head. It discusses a disagreement over rent payments between OpenAI and Neuralink — Smithies says “I’d expect [OpenAI] to get pretty nasty about it (ie probably willing to sue) if we didn’t pay something that they could point their auditors to,” saying “the main driver” is OpenAI accountants demanding it “so they can pass non-profit audits.” Birchall says he’ll “touch base with Chris to get his perspective.”
A January 2019 message chain between Musk and Birchall, concerning a reimbursement request from OpenAI for shared expenses with Neuralink in the Pioneer Building. Musk offers $250,000 and $1 million in payments for 2017 and 2018, respectively.
A full list of Elon Musk’s contributions to OpenAI, with entries dating from May of 2016 to September of 2020.
A copy of the Fidelity Charitable Policy Guidelines, 2017 to 2022.
An iMessage conversation from December 2024 between Musk and Meta CEO Mark Zuckerberg. Zuckerberg offers a “quick heads up that Meta sent a letter to the California AG supporting your lawsuit against OpenAI. Someone (not us) leaked leaked the letter and it will be public in the next hour. Wanted to make sure you heard this from me.” Musk replies: “Ok.”
A February 2025 iMessage conversation between Musk and Zuckerberg. “Are you open to the idea of bidding on the OpenAI IP with me and some others?” Musk asks. Zuckerberg asks to discuss live, and Musk says, “Will call in the morning.”
A letter from Musk’s xAI and several other investors to OpenAI, proposing an acquisition of all OpenAI’s assets.
Update, April 30th: Added newly available exhibits.
Update, May 1st: Added newly available exhibits.