文章探讨了人工智能领域资源分配不均的现象，作者将当前科技巨头垄断先进AI模型比作美国西部拓荒时代的终结，认为这标志着互联网这一"无许可创新空间"的消失，就像当年自由土地资源枯竭导致社会流动性下降一样，技术鸿沟正在扼杀普通人的发展机会。

边界闭合的时代——坦尼娅·维尔玛的忧虑

核心论点：
人工智能领域正重现19世纪美国"边疆关闭"的历史转折点，技术垄断将导致社会流动性固化。当Anthropic公司宣布其最先进的Mythos模型仅限企业合作伙伴使用时，作者首次真切体会到"被时代抛弃的贫穷感"。

历史隐喻：
文章援引历史学家特纳1893年提出的"边疆理论"——美国精神源于西部拓荒时代的自由土地，而边疆的消失标志着第一个美国时代的终结。如今数字边疆正在闭合：曾让16岁少年与亿万富翁共享同等技术起点的互联网平等性，正被资本主导的AI鸿沟取代。

垄断危机：
1. 资本霸权：如鲁道夫·莱恩所言，当AI替代劳动时，早期资本持有者将获得永久优势
2. 新封建主义：乔治·霍兹警告，智力垄断将创造永久"数字底层阶级"
3. 安全悖论：当前合作方（如经常发生安全事件的CrowdStrike）反而可能成为系统脆弱点

制度缺陷：
- 私营公司同时扮演立法者、执法者、司法者角色
- 缺乏公开准入标准、申诉机制和审计义务
- 安全研究被迫依赖中国开源模型（讽刺的是主流观点认为开源危险）

解决方案建议：
1. 建立类公共事业的准入制度，默认开放而非封闭
2. 设置可调整的安全护栏而非彻底封锁
3. 硬件发展可能使算力廉价化（如个人电脑超越大型机的历史）

深层隐喻：
作者将当前AI垄断比作殖民主义——提取全人类数据价值后，仅向内部小圈子分配收益，再用"AGI后的全民基本收入"等承诺作为道德遮羞布。文末以市政砍伐20年榕树防范飓风的荒诞案例，警示过度防范可能扼杀数字世界的生机。

（注：全文保留关键论据链，删减了部分重复性案例和技术术语解释，压缩了安全研究研讨会细节，突出"数字边疆闭合"的核心隐喻。）

评论内容总结

1. 关于Mythos模型的开放性与竞争

• 观点：部分评论认为Anthropic限制Mythos的开放是暂时的，竞争压力会迫使其最终公开模型。
  • "As long as there is someone somewhere in the world that releases a competing model then Anthropic will be forced to release Mythos." (pizzly)
  • "We are going to have access to this within 6 months, and if we don’t, someone else will offer an equivalent." (operatingthetan)
• 反驳观点：也有评论认为这是营销策略或资源限制。
  • "The ‘this model is too powerful for the general public’ sounds like marketing to me." (cortesoft)
  • "The problem is Anthropic doesn’t have the compute to deploy this model to scale to everyone yet." (OldGreenYodaGPT)

2. 开放模型与封闭模型的争议

• 支持开放：认为开放模型已足够好，封闭是特权化的表现。
  • "Open models are now good enough... The boasting of models so good they can’t be shared was propaganda." (lowbloodsugar)
  • "Intelligence should work similarly... the presumption should be access." (derektank)
• 支持限制：出于安全风险考虑，限制是必要的。
  • "It is sensible to believe that AI is reaching [the point where] we need to place restrictions on capabilities accessible to individuals." (fwipsy)
  • "All you need is one jailbreak from bad actors... it is highly risky to give it to the general public." (simianwords)

3. 技术与社会影响

• 技术乐观：认为技术进步会自然推动开放。
  • "This is now truer than it ever was... a 16-year-old with no credentials could just do things." (wyc)
• 社会批判：批评资源集中和权力垄断。
  • "The Internet was developed by the US state sector... Predictably driven by corporate interests." (keybored)
  • "The American dream was always a lie... there should not exist any superrich." (shevy-java)

4. 历史类比与未来预测

• 类比历史：部分评论用历史（如电力、美国西部开发）类比当前AI发展。
  • "Like electricity, AI access should scale with vetting but presume openness." (derektank)
  • "The early American frontier was a brief burst of unrestrained growth... individuals advance only during disruption." (fwipsy)
• 未来担忧：认为当前可能是公共AI访问的短暂窗口期。
  • "2019-2025 may be a rare blip in public AI access... incentives to offer public APIs are declining." (atleastoptimal)

5. 具体建议与质疑

• 建议：提出Anthropic应利用Mythos提供安全服务而非隐藏。
  • "Anthropic should attack businesses’ infrastructure using Mythos and issue vulnerability reports." (tim-projects)
• 质疑模型真实性：怀疑Mythos的实际能力或被夸大。
  • "My bet is this is just a model with slimmer safety alignment running 100 agents in parallel." (laurentiurad)

关键引用保留

• 开放与竞争：
  • "if it is an actual leap, expect more 90% is good enough open weights to hit the next 6 months." (cyanydeez)
  • "Anthropic hasn’t walked to the edge of the abyss only to be like ‘let the CEO’s handle this!’" (operatingthetan)
• 安全与风险：
  • "It’d only take one company deciding to not worry about safety, to change the calculus back to ‘we have to release this’." (hgoel)
  • "Eventually we reach the point where any malcontent can destroy the world by snapping their fingers." (fwipsy)