来自Claude Services的支持令人无法接受地糟糕。
我现在遇到了一些问题：
1.) 不知怎么的，我的账户状态变成了不再认为我有账户。每次我尝试登录时，它都会把我推到入门引导页面。在我的手机上，一切正常，并且我的专业计划也在。但在多个不同浏览器的隐私模式下，它就是强制我进行入门引导。我在这两种情况下使用的是完全相同的谷歌账户。
2.) 我下载了Claude桌面应用程序。尝试用我的谷歌邮箱登录时，它让我输入验证码。我查看了我的GMAIL，发现那不是验证码，而是浏览器会话的登录链接。哈哈，好吧。这根本就不行吗？这可以接受吗？
3.) 我没有办法与Anthropic的支持人员进行沟通。公司删除他们的支持系统并用代理取而代之，会导致人们对软件和质量的信任崩溃。我的情况确实需要人工处理。自动代理到最后会说我的问题不正常，然后把我重定向到我已经来过的常规支持链接。
如果我访问平台网站，它会正确显示我的账户信息，包括我设置的API密钥和组织信息，这些都很清楚。但当我尝试进入“设置->账单”时，它又把我重定向到入门引导页面。
在这种情况下，他们的状态机实际上是坏掉的。
看起来这个问题出现的时间正好与我购买OpenAI订阅以测试GPT 5.6的时间相近。我怀疑这不是阴谋，更像是一些合法的疏忽。
查看原文
Here's some issues I'm having right now:<p>1.) Somehow my account got into a state where it no longer thinks I have an account. Every time I try to login it just pushes me to on-boarding. On my mobile phone it works just fine and has my pro plan. But my browsers, in private mode in multiple different browsers, it just forces me to on-board. I'm using the exact same google account in both scenarios.<p>2.) I downloaded the Claude desktop app. Tried to sign in with my google email and it is asking me to enter a verification code. Go to my GMAIL. It's not a verification code, it's a sign-in link for the browser session. Lol. Okay. So that just literally doesn't work? Is this acceptable?<p>3.) There's no way to talk to a human about my support issue that works at Anthropic? Companies deleting their support systems and replacing them with agents is going to lead to the downfall of trust in software and quality. My situation literally needs a human. The automated agents get to the point where they say my issue is abnormal and just redirect me to the regular support link which I already came from.<p>If I go to the platform website it shows my account information properly even the API key and organization I setup so that all is clear. But then I try to go to say "Settings->Billing" It just redirects me to on-boarding.<p>Their state machine is actually broken in this scenario.<p>It seems like this issue appeared around the same time that I purchased an OpenAI subscription as well to test out GPT 5.6. I doubt it's conspiratorial here more like some legitimate carelessness.