FTX-Backed Anthropic’s CEO Declined OpenAI's Merger Offer: The Information

CoinDeskPolicy發佈於 2023-11-20更新於 2023-11-21

文章摘要

FTX bought a stake in Anthropic supposedly worth $500 million, according to an internal document circulated before last November’s bankruptcy filing.

FTX-backed artificial intelligence company Anthropic's CEO and co-founder Dario Amodei declined an offer from OpenAI's board of directors to merge, according to a report by The Information which cited a "person with direct knowledge."

The proposal came after OpenAI fired CEO Sam Altman on Friday and the deal had a sweetener – For Amodei to replace Altman as CEO, which he also turned down due to his position at Anthropic, according to the report.

"It’s not clear whether the merger proposal led to any serious discussion," the report said.

Advertisement
Advertisement

Later, Reuters also reported the offers were made and declined, citing two people briefed on the matter.

FTX bought a stake in Anthropic supposedly worth $500 million, according to an internal document circulated before last November’s bankruptcy filing. FTX's bankruptcy trustee has yet to sell the stake.

The AI firm's several fundraising efforts recently raised hopes that FTX creditors may receive a larger payout should the stake be sold. The U.S. Department of Justice alleges that the $500 million investment in Anthropic in 2022 came from customer funds.

Edited by Omkar Godbole.



你可能也喜歡

大神Karpathy用Claude的方式,原来是这样的?

AI大神安德烈·卡帕西(Andrej Karpathy)在加入Anthropic后,开源活动减少。近期,一份据称是他本人使用的`CLAUDE.md`文件在社区流传,旨在指导Claude AI进行更高效的编程协作。该文件强调了一系列核心原则,并非简单建议而是必须遵守的规则,以规避大语言模型写代码时的常见错误。 核心原则包括: 1. **写之前先读**:在编写新代码前,务必认真阅读现有代码库,理解项目风格、既有模式和工具,确保新代码与项目保持一致。 2. **写代码之前先想清楚**:明确任务需求、澄清假设、说明方案取舍,避免基于模糊理解生成错误代码。 3. **保持简单**:抵制过度设计,避免过早抽象、臆想式错误处理和不必要的配置,编写能解决当前问题的最少代码。 4. **外科手术式修改**:修改代码时力求改动范围最小,严格匹配项目现有风格,只清理自己引入的问题,避免无谓的重构和格式化。 5. **验证**:重视测试,修复bug前先写测试复现,确保改动不破坏现有功能。 6. **目标驱动执行**:将模糊任务转化为可验证的具体步骤,并在执行前说明计划。 7. **调试**:遇到问题先深入调查、复现,理解根因,避免盲目添加临时解决方案。 8. **谨慎管理依赖**:添加新依赖前,优先考虑使用现有工具或标准库,并评估其维护状态和成本。 9. **有效沟通**:清晰说明改动内容和原因,主动指出潜在隐患和不确定之处。 文章指出,尽管这份文件的真实性存疑,但其内容高度契合卡帕西本人关于AI编程助手的公开观点。社区已据此提炼出原则并制作了模板,有测试表明能显著降低Claude的代码错误率。这些准则被认为是提升AI辅助编程效率、保证代码质量的关键。

marsbit1 小時前

大神Karpathy用Claude的方式,原来是这样的?

marsbit1 小時前

交易

現貨
活动图片