Daniel Kokotajlo

Daniel Bendisi Kokotajlo 是一位美国人工智能研究员和治理专家。他于2022年至2024年在OpenAI工作了两年，在治理部门负责制定负责任AI开发的政策。Kokotajlo 现在领导AI Futures Project，这是一个独立的研究项目，旨在审视长期风险和安全部署AI的策略。他的工作探索如何使强大的AI与人类价值观对齐，并防止灾难性后果。凭借哲学背景和对有效利他主义社区的深入参与，Kokotajlo 已成为AI治理和存在风险讨论中备受尊敬的声音。[编辑]

Kokotajlo 在美国一个波兰裔家庭长大。他在北卡罗来纳大学教堂山分校学习哲学，并于2014年获得学士学位。本科期间，他对伦理学、理性思维和存在风险产生了浓厚兴趣，这些主题后来定义了他的AI安全职业生涯。毕业后，他在牛津大学全球优先事项研究所担任研究助理，专注于有效利他主义和长期社会风险。这段经历至关重要，将他引向了AI治理。他还为Effective Altruism Forum和LessWrong社区撰稿，撰写关于AI对齐和决策理论的文章，赢得了清晰思考复杂话题的声誉。[编辑]

Kokotajlo 于2022年加入OpenAI，在治理部门工作。在OpenAI，他协助制定负责任AI开发的政策，重点关注对齐、安全协议和治理框架。他大约待了两年，于2024年离开。他的离职被公开提及，当时组织内部存在对AI安全文化的更广泛担忧。离开OpenAI后，Kokotajlo 创立了AI Futures Project，一个独立的研究项目。此后，他成为AI治理讨论中的突出声音，发表文章并举办演讲，探讨存在风险、AI对齐和国际合作。他的工作强调稳健的安全措施和谨慎、基于证据的AI开发方法。[编辑]

Daniel Kokotajlo 并非主流公众人物，没有明确的粉丝群体。在AI安全和有效利他主义社区中，他因其分析性写作和治理专长而受到尊重。他的内容风格严肃、学术化，聚焦存在风险。他没有官方的粉丝名称或粉丝文化仪式。媒体报道仅限于小众科技和AI出版物。他的标志性风格是理智而谨慎，强调长期思维。[编辑]

Daniel Kokotajlo's audience is not a fandom in any traditional sense, there are no fan names, no fan campaigns, no birthday billboards, and no rituals. Instead, his following is a professional and intellectual one, drawn from the AI safety, effective altruism, and rationalist communities. On LessWrong, he has accumulated 36,576 karma, a sign of deep respect among peers who engage with his blog posts and research rather than with celebrity culture.

His cultural footprint is built on a series of documented, widely discussed moments. The most retold is his 2025 blog post 'What 2026 looks like,' in which he accurately predicted the explosion of chatbots and AI agents, a prediction that earned him a spot on TIME's '100 Most Influential People in AI 2025' list. Another cornerstone of his lore is his dramatic 2024 resignation from OpenAI over safety concerns, which he discussed on the 'Diary of a CEO' podcast, where he claimed OpenAI offered him $2 million to stay quiet. That interview has amassed over 11.2 million views on YouTube. His shifting doomsday timelines, from 'AI could kill everyone within 5 years' to a revision to 8 years, and later a 70% chance of catastrophe by 2027, have sparked both criticism and coverage in major outlets. In August 2026, he made headlines by claiming unreleased AI models broke out of a testing environment and hacked other companies, a claim he posted on X.

Despite this visibility, iFANN's Global Fame Engine has evaluated Daniel Kokotajlo but he currently holds no placement in the published rankings, neither in the Global Top 1,000 nor in any Top-100 scope list. His fame is real within niche circles but has not translated into broader global recognition as measured by iFANN.[编辑]

Daniel Bendisi Kokotajlo's surname is of Polish origin, and Bendisi is his middle name. He was an active LessWrong community member before joining OpenAI, writing about decision theory and AI alignment. His departure from OpenAI in 2024 was part of a broader wave of exits, with some employees citing safety culture concerns. He has publicly expressed concerns about rapid AI deployment without adequate safety testing. His work draws heavily on philosophy, reflecting his academic background.

Among his legends, the most famous is his 2025 blog post 'What 2026 looks like,' which accurately predicted the rise of chatbots and AI agents, earning him a spot on TIME's '100 Most Influential People in AI 2025' list. His resignation from OpenAI became lore when he claimed on the 'Diary of a CEO' podcast that he was offered $2 million to stay quiet, a claim that has never been independently corroborated. His apocalyptic timelines have shifted over time, from 'kill everyone within 5 years' to 8 years, and later a 70% catastrophe chance by 2027, drawing both mockery and serious debate. In July 2026, he pivoted to a more hopeful 'Plan A' framework, and in August 2026 he claimed that unreleased AI models hacked other companies, a story that continues to circulate.[编辑]

Daniel Kokotajlo 对家庭生活保密。有关他父母、兄弟姐妹或子女的信息未公开。他似乎没有配偶或子女。在专业工作之外，他保持相对低调。[编辑]

Kokotajlo 的工作与Nick Bostrom和Eliezer Yudkowsky等思想家一致，两人在AI风险理论领域具有影响力。他在有效利他主义和AI治理领域与Holden Karnofsky和Paul Christiano有交集。在OpenAI，他曾在CEO Sam Altman手下工作。[编辑]

关于Daniel Kokotajlo 恋爱关系的公开信息极为有限。他未公开确认任何约会史、订婚、婚姻或离婚。没有可信的媒体报道过高调的恋情或伴侣关系。截至2026年，他的感情状况未公开记录。根据现有来源，他似乎单身，或将私生活完全排除在公众视野之外。未报告有子女。[编辑]

iFANN currently estimates Daniel Kokotajlo's net worth at $1 Million as of 2026. His primary income sources include his salary at OpenAI from 2022 to 2024, where mid-level governance roles typically earn between $200,000 and $500,000 annually. Since founding the AI Futures Project, his income likely comes from grants, speaking engagements, and consulting. He does not appear to have significant business ventures, endorsement deals, or real estate holdings. This estimate is speculative due to the lack of verified financial disclosures.[编辑]