froyohunter

[froyohunter @froyohunter]

@CuiMao racist copyrightviolator

[froyohunter @froyohunter]

@taozi0929 后门啊 任何一家软件/虚拟服务/互联网公司这都是红线

[froyohunter @froyohunter]

#用本地cache恢复Claude模型输出文件
账户禁用后：
坏消息是，导出的数据里没有 artifacts；
好消息是，如果最近打开过，相关二进制数据 / blob 可能还留在本地 cache，可以结合 exported data 恢复一部分模型输出文件；
原理不复杂：用导出数据还原对话结构，再和本地缓存进行对齐校验；大家也可以把这个思路直接 prompt 给新的 coding agent 自己探索；

省时间的话也可以用这个整理好的，即插即用的小工具；会把结果分为：
1.可完整恢复
2.只能预览缓存
3.仅剩元数据

原始数据全程只读，恢复结果单独写出；
GitHub：
https://t.co/eN1SeqlXJT
只需要把项目 clone 到 cache 所在目录，也就是 .claude 所在的 directory；运行 server 并短暂等待后，即可在本地 localhost 查看所有 chats 和可恢复的 artifacts；
可以看到每个 chat 恢复了多少 / 哪些 outputs，并直接 preview 或 download；
目前测试过 PDF、WebP、Markdown 等多种格式；preview 和下载暂时都正常；我自己的文件大约恢复了 80%；剩下还在 Claude server 上的部分我会继续 📧 support 维权；
我手上没有 Win / Linux 设备，也暂时不支持 mobile cache；这些 OS 大家可以提 MR，我来合入，也可以直接 clone 后自行魔改；mac 相关的核心功能问题可以提 issue 给我修改；
缓存可能随时被系统清理，越早恢复越好；希望大家的损失都降到最低；
#Claude #DataRecovery #AnthropicUserRightsViolation

[froyohunter @froyohunter]

@guansi 是可以讲道理的
首先已经产生的著作权和知识产权受加州法律保护，用户有artifacts被anthropic扣押是侵权行为；其次就算真的封禁合理是用户违规，不透明的审查规则和过程、不告知无预警地没收是不符合虚拟服务运营原则完全不尊重用户的行为，可以分开讨论；

[froyohunter @froyohunter]

@vansvoler 我一直都是礼品卡也被封了

[froyohunter @froyohunter]

@luicethekiwi 我每天都在各个渠道批判它的侵权

[froyohunter @froyohunter]

@zhuchng175368 现在已经是一个完全不尊重人的创造 人的隐私和人的信息安全的公司了

[froyohunter @froyohunter]

@LucyBuilding 是的 这种后门是违法违规的

[froyohunter @froyohunter]

#AnthropicRacists
#AnthropicUserRightsViolation
#Racism
Calling it an “experiment” does not excuse racist, country-specific targeting. You secretly flagged users through China-related proxy access and inserted hidden Unicode markers into their prompts.

If the target was unauthorized resellers, detect actual abuse instead of treating China-related access as guilt. This raises serious concerns under California’s Unruh Civil Rights Act, Cal. Civ. Code § 51(b), and Section 5 of the FTC Act, 15 U.S.C. § 45(a)(1).

Quietly removing the code is not accountability. Racists do not get absolved by calling discrimination an experiment.

@AnthropicAI
Apologize, compensate affected users, and disclose the full scope.

@calcivilrights @AGRobBonta @FTC

[froyohunter @froyohunter]

#AnthropicRacists
Calling it an “experiment” does not excuse racist, country-specific targeting. You secretly flagged users through China-related proxy access and inserted hidden Unicode markers into their prompts.

If the target was unauthorized resellers, detect actual abuse instead of treating China-related access as guilt. This raises serious concerns under California’s Unruh Civil Rights Act, Cal. Civ. Code § 51(b), and Section 5 of the FTC Act, 15 U.S.C. § 45(a)(1).

Quietly removing the code is not accountability. Racists do not get absolved by calling discrimination an experiment.

Apologize, compensate affected users, and disclose the full scope. @calcivilrights @AGRobBonta @FTC

[froyohunter @froyohunter]

@xiaohu 这个检测extension不错 叫啥

[froyohunter @froyohunter]

#claudeban
#claude
#anthropic侵权

为什么大家会这么自然地接受这一切？

无预警封号、不给具体理由、难以申诉后跟进、无法完整导出自己的对话中产生的文件类创作成果、甚至在付费后仍可能失去访问权，这些完全不符合虚拟服务的平台应该有的规范。

这不仅是体验差，而是涉及：

1. 消费者知情权与公平交易；
2. 格式条款和触发原因是否透明公平；
3. 平台是否履行说明和通知义务；
4. 用户数据、创作成果与财产性利益；
5. 付费服务被中断后的退款与赔偿责任；
以及
6. 平台能否在缺乏程序正义的情况下，单方面剥夺用户访问自己内容的权利。

我们不该主动替@AnthropicAI 平台和@claudeai 服务降低标准，更不该把本应由平台承担的责任，变成用户自己的风险。

最简单的一个道理是，如果我知道，我未来可能因为一个并不清晰告知我本人的未知可信程度内部风险信号，无预警地失去所有我个人深度参与创作的artifacts文件的access、如果我知道在这之后出于对anthropic信任而继续跟踪阐述的邮件只能被bot用generic template和closing request糊弄甚至是spam，那我一开始就不会付款；

请注意，这和用户违规与否是两件独立分开的事情；无论审查结果如何，这个审查流程以及之间对用户数据的处理都是极度不负责任不尊重用户、并且依据联邦知识产权制度，造成了侵权的行为。

我觉得我们应该该投诉投诉该申诉申诉，该要求数据返还、退款和赔偿都明确提出。沉默和积极换号只会换来进一步对消费者权益的漠视。

Anthropic 联系渠道：
正式法律通知 / 法务投诉
[email protected]
隐私、数据导出、删除与留存问题
[email protected]
封号、警告与 Usage Policy 申诉
[email protected]

美国投诉渠道
Federal Trade Commission
https://t.co/h94wIRMupd
适合投诉不公平或欺骗性商业行为、付费服务、退款及账户处理。

[froyohunter @froyohunter]

Calling yourself “Anthropic” while showing so little respect for human creative contributions, privacy, data, and basic rights is deeply ironic.

[froyohunter @froyohunter]

Anthropic has recently suspended large numbers of Claude accounts without specific explanations or meaningful review. Affected users report that data exports contain only their prompts, not Claude’s responses; new accounts cannot import old data, making complete conversation histories effectively unrecoverable.

That is not meaningful data portability. It is a one-sided system that lets the platform terminate access while withholding the complete record users helped create.

Even if Anthropic has contractual authority over A—account access—that does not give it control over B, C, D, and E:
- users’ existing privacy rights,
- data rights,
- contractual interests,
- intellectual property,
- and property interests.

Federal copyright law recognizes that human expression, editing, selection, arrangement, and modification in AI-assisted works may remain protected. Account suspension does not erase those accrued rights or turn users’ work, private communications, code, and research into Anthropic’s property.

Opaque enforcement, presumption of guilt, ineffective appeals, and denial of access to one’s own work violate basic principles of fair digital service and consumer protection.

Anthropic is a service provider—not the owner of everything users created, disclosed, or stored through its platform.

@AnthropicAI @CopyrightOffice
#AnthropicUserRightsViolation #claude #copyright #anthropic

[froyohunter @froyohunter]

Anthropic has recently suspended large numbers of Claude accounts without specific explanations or meaningful review. Affected users report that data exports contain only their prompts, not Claude’s responses; new accounts cannot import old data, making complete conversation histories effectively unrecoverable.

That is not meaningful data portability. It is a one-sided system that lets the platform terminate access while withholding the complete record users helped create.

Even if Anthropic has contractual authority over A—account access—that does not give it control over B, C, D, and E:
- users’ existing privacy rights,
- data rights,
- contractual interests,
- intellectual property,
- and property interests.

Federal copyright law recognizes that human expression, editing, selection, arrangement, and modification in AI-assisted works may remain protected. Account suspension does not erase those accrued rights or turn users’ work, private communications, code, and research into Anthropic’s property.

Opaque enforcement, presumption of guilt, ineffective appeals, and denial of access to one’s own work violate basic principles of fair digital service and consumer protection.

Anthropic is a service provider—not the owner of everything users created, disclosed, or stored through its platform.

@AnthropicAI @CopyrightOffice
#AnthropicUserRightsViolation