Ask HN: 问 HN：我们都忘记了 AI 公司做过的哪些坏事？
1 分•作者: Imustaskforhelp•4 个月前
我最近在写评论时，突然意识到 GPT 5 视频里的图表有多糟糕。我差点都忘了这事。<p>我想创建一个非常小的网站，专门讨论这些问题。我觉得这些公司似乎存在一些扭曲的激励机制，它们只展示好的一面，隐藏坏的一面，而我们的新闻周期却不断更迭，我们所有人都在追逐着它。<p>此外，还可以列出一些值得关注的事件。<p>以下是一些例子：<p>OpenAI 从非营利组织转变为开放组织。<p>Claude 代码软件泄露事件（Anthropic）。<p>X/Grok 生成非自愿深度伪造的问题（XAI）。<p>OpenAI 与军方的交易，以及背叛行为，现在又推动制定新法律。<p>OpenAI 的图表问题，正如我之前提到的，实在太糟糕了。<p>Suchir balaji 的去世（我强烈建议大家搜索一下，了解一下背景）。<p>微软的 Co-pilot 相关问题。<p>Anthropic 发现中国模型从它们那里提取信息后，显示它们可以查看 API 中发送的内容，并且有一些相关的评论，我忘记了，但听起来对隐私不太友好。<p>OpenAI、Oracle 等公司的星际之门项目，导致了最初的内存危机。<p>值得关注的事件列表包括：<p>OpenAI 的 Sam Altman 公开结巴，并说出了现在极具标志性的一句话：“想卖掉你的股份吗？我会帮你找到买家”。<p>人们提到这些 AI 公司减少了人们可以发送的消息数量，以及所有关于这些模型变得越来越笨，但仍然收取 200 美元、100 美元费用的评论。<p>微软已经将 75 个产品命名为 Co-pilot，我想都是。<p>Anthropic 以一种非常病态的方式对待他们的 Mythos 模型，就像对待自己的孩子一样。<p>Deepseek 在其网站首次发布时，如果你说习近平想要这项工作，它会优先给你答案（推测）。<p>我希望大家能提供更多例子，我想整理一个列表，并围绕这个问题进行讨论。感谢您的阅读，祝您有个愉快的一天，期待在这里阅读评论！
查看原文
I was writing a comment recently when I realized just how bad the graphs in GPT 5 video are. I had almost forgotten about it.<p>I wish to create a very minor website which can talk about all of these, I feel like there are some perverse incentives which make these companies show out the good and hide out the bad and our news cycle just keeps on moving and we all follow it.<p>Also there can be some remarkable mentions as well which I wish to list<p>Here are some other examples:<p>OpenAI turning from non profit to open.<p>Claude code software getting leaked (Anthropic)<p>X/Grok generating non consensual deepfakes issue (XAI)<p>OpenAI's deal with Military and backstabbing basically and now pushing for new laws.<p>The graph thing by OpenAI being so horrendous which I mentioned earlier.<p>Suchir balaji's death (I genuinely recommend searching this up if someone hasn't for context)<p>Microsoft Co-pilot things<p>Anthropic when detected that chinese models distill from them showed that they can view things sent in API and there were some comments related to it which I forgot but it doesn't sound privacy friendly.<p>The stargate project by OpenAI,Oracle etc. leading to the ram crisis in the first place.<p>Remarkable mentions list includes:<p>OpenAI's Sam altman visible stumbled and saying now iconic line ‘Want to sell your shares? I’ll find you a buyer'<p>People mentioning how these AI companies have reduced the amount of messages people can send and also all the comments about the dumbing down of these models while still charging 200$ 100$ for these things.<p>Microsoft has named 75 products I think all on Co-pilot<p>Anthropic treating their mythos model like their own baby in a very psychotic way.<p>Deepseek giving you answers in high priority in their website when it was first released if you said that Xi Jinping wanted the work (speculation-ish)<p>I would love if people can give more and more examples about it and I wish to compile a list and have a discourse surrounding this. Thanks for reading this and have a nice day and waiting to read the comments here!