周锋锁 Fengsuo Zhou

人道中国@hchina89 创办人, Tiananmen leader 1989, 中国人权执行主任 Executive Director of @hrichina 个人观点 Telegram 订阅频道

[周锋锁 Fengsuo Zhou @ZhouFengSuo]

六四35周年纪念，我在美国国会听证会上的发言。得到普遍支持的八九民运展示了中国人要求自由民主的梦想，六四屠杀展示了中共政权的邪恶残暴，杜绝了中共自新改变的可能。美国和民主国家必须把中国民主化作为对中共政权政策的基本原则，否则就是自我毁灭。八九年中国曾经离自由一步之遥，美国也应该全力支持自由民主中国的愿景，支持为自由民主人权法制而奋斗的中国人，而不是绥靖专制政权。

[XiaojuZhang(原推陣亡)佐羅🇭🇰🇹🇼🇺🇸🇺🇦 @DemEnthusiast2]

#CDEF
#中國民主教育基金會

第四十屆「中國傑出民主人士獎」評選工作正在紧锣密鼓的筹备中。

今年，我們特別開放部分評選權重，邀請網路意見領袖及民運人士共同參與提名。

誠摯邀請您推薦心目中的合適人選，或協助轉發提名資訊。

為避免重複提名，敬請先參閱歷屆得獎名單，凡已獲獎者，請勿再次提名。

歷屆得獎名單：
https://t.co/DiIAPQlMkh

非常感謝您的支持與參與！
有意提名者，請發私信。

[Superman @thesupermannx]

Researchers proved every LLM is secretly obsessed with Japan.

And they finally figured out why.

They published a paper investigating hidden cultural biases in large language models.

they tested 31,000 questions across 24 different languages on models like gemini, chatgpt, claude, and llama.

they stripped out all geographic hints and just asked the models to pick a place and answer.

the finding was universal.. across a wide range of models and languages, there is a pervasive "obsession" with japanese culture.

even when you prompt them in a language with zero ties to japan, they default to japanese norms, values, and aesthetics at a disproportionately high rate compared to its actual global demographic weight.

the craziest part? they tracked down exactly when the ai develops this obsession.. and it is not from reading the raw internet.

during pre-training, the models actually have a fairly balanced, global view of the world. the obsession gets injected during the safety phase.

when tech companies fine-tune these models to be "helpful, harmless, and polite," the ai starts mapping those corporate safety guidelines directly onto japanese cultural concepts. in the pursuit of creating the perfect, inoffensive, compliant assistant, they warped how the models view the world.

the quantization process (compressing models to run faster locally) actually amplifies this determinism. the smaller and faster your local model is, the more stereotypically it behaves, causing japanese cultural tropes to spike hard.

we spend all this time worrying about existential ai safety, but the real alignment problem is that our most advanced reasoning engines are hyper-fixated on japanese internet culture.

every model upgrade is reinforcing an invisible regional bias that dictates how they generate stories, facts, and code. the cultural commonsense of the internet is not universal..

it’s just localized to whatever the heaviest reddit and anime wiki users were obsessed with.

[伍雷（吊证律师） @lijinxinglvshi]

【号外】三个水枪手官方电报群已经建立了。现在，发布电报群二维码，欢迎来自全球各地苦秦久矣的朋友们，扫码加入！

[中国人权-Human Rights in China @hrichina]

【因讽刺习近平是“第一傻X”，台湾相声演员黄逸豪先后被泼粪及买凶撞伤】台湾相声演员黄逸豪因一段讽刺习近平的相声短视频，先后遭遇泼粪和买凶制造假车祸。新北地检署近日已起诉5名涉案嫌犯，案件被认定为境外人士出资的跨境镇压。据台湾镜周刊报道，黄逸豪去年12月发布相声短视频，以“独辱习不如众辱习”为题讽刺习近平。今年5月他加场演出时，有人受雇前往场地泼粪。9月11日下午，他在新北市永和区住处附近骑机车停车时，被一辆汽车从后方猛撞。监视器画面显示，肇事司机直到听见他的惨叫才停车，另有一名共犯一直守在巷口监视，并不时发信息向同伙回报。

调查显示，出资者是一名代号“K”的中国籍人士。事发后，嫌犯又以协助处理理赔为由，骗取黄逸豪载有病历和身份资料的诊断证明书，翻拍后层层转交给K，用来领取酬劳。K因黄逸豪只是骨盆受伤，未达到预定的“断手断脚”，还扣减了尾款。

黄逸豪今天公开受访，称被撞倒后的第一反应是“该不会是老共来找我了吧”。肇事者下车后姿态放得很低，不断道歉，还热情邀请他坐进车内休息。他无法确认对方来路，坚决拒绝。他还说，对方花了新台币45万元，而他那场演出的总票房只有13万元。

以“断手断脚”为目标，只为惩罚一段讽刺领导人的相声，这早已超出言论分歧的范畴。跨境镇压正从网络围攻和骚扰，发展为直接的人身伤害，而且发生在台湾的街头。言论自由不应以身体安全为代价，对中共这种肆无忌惮的暴力跨境镇压，各国政府和社会都不应视而不见。

[多伦多方脸 @torontobigface]

加拿大不愧是民主制度的典范
最近的阿尔伯塔省独立，公投就投了，没有人因为对方要闹独立
就要把对方打成罪犯，要杀留阿省不留人
符合程序就进行投票

现在投完了，没有通过，也就没下文了

大家都有在宪法框架下表达自己想法的权利，也都有愿赌服输的精神

[中国人权-Human Rights in China @hrichina]

#中国人权 #每日一推 #良心犯 第497期:今天(2026年10月20日)获刑6年半的香港47人案之郑达鸿狱中度过38岁生日。因曾签署反对香港国安法的声明、争取外国制裁、否决港府财政预算案等。2021年初在民主派初选大搜捕中遭国安处拘捕,后被香港高院判"串谋颠覆国家政权"罪成,被定为"反中乱港分子",其庭上不认罪

#郑达鸿：英文名 Tat Cheng Tat-hung，1988年10月20日出生，香港岛炮台山居民，泛民主派人士，前公民党青年公民主席、公民党副秘书长及执行委员会成员，前香港东区丹拿选区议员，北炮同盟（已解散）成员。被中共定为"反中乱港分子"，香港泛民主派人士。

郑达鸿自12岁起在香港岛东区北角长大，先后就读张祝珊英文中学、圣士提反书院及香港理工大学，取得社会政策及行政荣誉学士学位。

2011年郑达鸿成为公民党九龙西支部实习生，毕业后正式加入公民党，并在北角开展地区工作。

2013年起，郑达鸿在丹拿选区服务，关注北角海滨规划、地产发展、官商关系、棕地发展、丁屋政策及皇都戏院保育等议题，并多次参与城市论坛。

2014年至2016年，郑达鸿担任公民党青年公民主席，亦曾任港岛支部副主席。

2015年，郑达鸿当选东区丹拿选区议员，成为该选区首位民主派区议员；

2016年至2018年郑达鸿担任公民党副秘书长，并取得香港城市大学法律博士学位。

2019年区议会选举期间，郑达鸿与其他参选人组成“北炮同盟”，参与北角、炮台山多个选区的选举。

2020年，郑达鸿代表公民党参加香港立法会选举民主派初选，7月22日以港岛区第四高票胜出；但8天后，其因多项议题的政治立场与态度被询问，包括涉及其曾签署反对港区国安法的声明、争取外国制裁、否决财政预算案等范畴，遂其竟被选举主任裁定为提名无效；同年12月15日，其在社交网站上宣布退出公民党；。

2021年1月6日，郑达鸿在香港民主派初选大搜捕中被警方国安处拘捕，3月13日获保释；同年5月21日，其等12名各区民主派议员被曝请辞，之后其又被收押候审。

2024年5月30日，香港高等法院3名国安法指定法官裁定其等47案被告「串谋颠覆国家政权」罪成。

2024年11月19日，其等"47人初选颠覆罪"案被告在西九龙裁判法院统一宣判，其因"反中乱港分子"和"不认罪"而被判囚6年6个月（78个月），刑期预至2030年。

谨此祝福郑达鸿狱中平安,生日快乐,早获自由!
#HRIC #关注在押良心犯 #郑达鸿 #香港47人案

[周锋锁 Fengsuo Zhou @ZhouFengSuo]

黄逸豪万幸。中共国安企图遥控制造车祸攻击熊焱，在2023年被FBI曝光，与此高度类似，不知是否同一人，或者同一机构。

[王剑 Wong Kim @wongkim728]

又见跨境镇压！
台湾相声演员黄逸豪因为公开讽刺习近平，竟然遭人买凶制造车祸。新北地检署10月21日起诉五名嫌犯。检方查出，一名代号“K”的中国人士涉嫌出资45万新台币，指使人员跟踪黄逸豪、安装GPS定位器，企图开车撞断他的手脚，阻止他登台演出。9月11日，黄逸豪遭蓄意冲撞，幸好仅受挫伤。境外金主还嫌他伤得太轻，拒绝支付20万元尾款。

[王剑 Wong Kim @wongkim728]

又见跨境镇压！
台湾相声演员黄逸豪因为公开讽刺习近平，竟然遭人买凶制造车祸。新北地检署10月21日起诉五名嫌犯。检方查出，一名代号“K”的中国人士涉嫌出资45万新台币，指使人员跟踪黄逸豪、安装GPS定位器，企图开车撞断他的手脚，阻止他登台演出。9月11日，黄逸豪遭蓄意冲撞，幸好仅受挫伤。境外金主还嫌他伤得太轻，拒绝支付20万元尾款。

[中国人权-Human Rights in China @hrichina]

Recommended Reading: What Can Democratic Forces Do in the Age of AI?

The October 2026 issue of China Democracy Quarterly features a roundtable, "Democratic Opportunities in the Age of Artificial Intelligence," with five scholars and practitioners: Genia Kostka, Xiao Qiang, Alex Gladstein, Jiang Zhiqiang (季志强), and Liu Shuang (刘爽). The question at hand: as AI hands authoritarian regimes new tools of control, does it also open new space for democratic forces?

Genia Kostka (Professor of Chinese Politics, Freie Universität Berlin) argues that the binary of "democracies use AI to improve public services, authoritarian states use it to surveil citizens" is increasingly inadequate. Both types of regimes use many of the same technologies, for administration and for social control alike. In her view, AI's democratic potential lies not in the technology itself but in the institutions and social practices surrounding it, and the central contest is over who can see, challenge, and constrain the ever-expanding power of the state.

Xiao Qiang (Founder and Editor-in-Chief, China Digital Times) introduces the concept of an "intelligence tax": once censorship must reach into a large model's training, alignment, and generation, it comes at a cost and leaves cracks visible from outside. He identifies memory, correlation, and scale as three levers democratic forces can draw on, but stresses that "cracks do not automatically become levers." The key is to use the window while it lasts to build knowledge infrastructure that endures even after the cracks close.

Alex Gladstein (Chief Strategy Officer, Human Rights Foundation) argues that AI agents are a direct capability multiplier for chronically underfunded China human rights organizations, especially exile communities and activists working on Taiwan, Hong Kong, Tibetan, and Uyghur issues. Open-weight models that can be run locally and customized, he says, pose the first real technical challenge to the CCP's near-total control of the digital space over the past decade. He also notes that the frontier open-weight models currently usable in confidential computing environments come mainly from Chinese labs.

Jiang Zhiqiang (季志强; PhD in political philosophy, machine learning engineer, and computational politics researcher) approaches the question from freedom of expression. Even if model development is concentrated in a few companies and governments tighten control, ordinary people can still use these tools to expand their own capabilities. AI lowers the barriers to both accessing information and expressing and spreading it, from assisting verification and cross-language translation to producing content and connecting activists across countries. But models make mistakes, and their value lies in aiding verification, not in becoming a new arbiter of truth.

Liu Shuang (刘爽; independent commentator and AI writing experimenter) dissects the authoritarian system's inherent contradictions at the technical level: officials demand that models refuse sensitive political questions at a very high rate, yet also require them to answer "politically correct" on grand narratives, a standard that undermines itself. She also warns that users inside China probing sensitive topics with domestic chatbots face serious risk, and that caution remains the first rule. In her conclusion, she writes that every deleted post recorded and every erased name preserved today plants a seed of truth for tomorrow's models.

Technology does not bring democracy automatically. But while the cracks remain, documentation and preservation are still worth doing.

Full text 👇
https://t.co/Ijl8ToN4C9

[Fear_Nation_世界苦茶 @Ansel_Flipradio]

最近臺灣地方選舉因沈伯洋大熱。導致x出現一波反民主的熱潮，認為中國一黨專制更優秀。因為選舉，很多人尤其對比票選制和“選賢任能”。
認真的嗎？中共這些官員的言行，從習近平到下面，和賢能有一分錢關係嗎？這些人的威風和能力，都是背靠一個14億人的汲取系統，離開這套吸血體系，他們什麼都不是。

[中国人权-Human Rights in China @hrichina]

The home of a Dalai Lama representative in South Africa was firebombed on October 14; Feng Rui (Xiao Wu) has reportedly fallen ill from interrogation; imprisoned pastor Wang Yi needs medical care for advanced liver cancer; and more in the Weekly Brief:
https://t.co/7AnKdpxbz7

[周锋锁 Fengsuo Zhou @ZhouFengSuo]

六賢臺上山紫水明，四聖堂前月白風清。六四，图二的对联令我心有戚戚焉

[周锋锁 Fengsuo Zhou @ZhouFengSuo]

似曾相识我归来，东京的哲学堂公园

[周锋锁 Fengsuo Zhou @ZhouFengSuo]

似曾相识我归来，东京的哲学堂公园

[Students For Liberty @sfliberty]

This is a guy that many young Americans look up to...

[National Review @NRO]

October is known for disgusting political attacks, but we cannot be indifferent or tolerant -- this must be condemned. | @kathrynlopez

https://t.co/3eKvvRIMyk

[美国之音中文网 @VOAChinese]

“第一批子弹射在我们前面。”37年前，29岁的查建英亲眼目睹六四开枪：不断有人倒下，马上有人上前救援。上千人怒喊“畜生！畜生”她仍记得白衬衫上的鲜血，如“黑花在绽放”。
《原声带：查建英（上）｜六四，我在广场·看这个政权再次露出獠牙》完整专访：美东10月24日晚8点

[Dajun Zhang @ZhangDajun]

华人移民一直都是优秀的。华人应该对自己的身份感到自豪。

下面对巴黎郊区华人移民和其他族裔移民的对比，可以佐证我的观点。

[中国人权-Human Rights in China @hrichina]

【中共强推社保覆盖惹舆论反弹】据路透社，中国正在加大力度推动社保扩面征缴：人社部副部长李忠表示，要“想方设法”追缴未缴费人员并动员其参保，包括借助大数据手段；11月1日起，对少缴漏缴社保的用人单位将统一处罚标准，最高可罚欠缴金额的三倍。

政策一出，舆论反应强烈。微博话题“年轻人为什么不愿意交社保”周一登上热搜榜首，阅读量至少4800万，到周二已显示“根据相关法律法规和政策，该话题不予显示”，但相关帖子和评论仍可看到。另一个关于用人单位欠缴社保新罚则的话题，阅读量也超过2150万。

网民的诉求相当一致：先把工资和工作岗位解决好。有人说，自己的收入只够维持生活，实在拿不出钱来缴；有人说，帮大家找到有社保的好工作，问题自然就解决了；还有人干脆留言，让机器人来缴社保。眼下，员工个人缴费约占工资总额的10%，用人单位约占25%，在房地产危机持续、收入增长乏力的背景下，这对普通家庭和中小企业都是不小的负担。

数据也说明了问题的另一面：2025年一项调查显示，只有34.1%的受访企业称社保缴纳完全合规；超过3亿灵活就业人员本就不在强制缴费之列，多数人选择不缴全部险种。

当局希望通过完善社保来减少居民的预防性储蓄、刺激消费，这一思路本身并不难理解。但在就业市场持续走弱、民众对养老金能否兑现心存疑虑的情况下，先加码征缴、后谈保障，难免被看作把压力转嫁给了本就吃紧的劳动者。信任不是靠罚款和大数据追出来的，如果工资和就业这个前提迟迟不能改善，类似的民怨还会不断出现。

报道原文👇
https://t.co/09whHHonqh

[China Uncensored @ChinaUncensored]

It’s amazing how quickly rape becomes permissible to these people when it’s done in the name of Revolution.

[Visegrád 24 @visegrad24]

🇸🇦🇧🇩 Saudi Arabia executed two Bangladeshi nationals, Nazrul Islam Abdul Basir and Siti Akhtar, for what authorities called an “immoral” or “indecent act” inside a mosque in Mecca.

The Interior Ministry said a court handed down the death sentences under ta’zir, which allows judges broad discretion where fixed penalties do not exist.

The court deemed the sentence "appropriate" because of the "severity of violating a holy site." The verdicts were upheld and finalized by royal order.

Viral claims online said the couple were newlyweds executed for a forehead kiss. Other posts claimed they were caught in an explicit sexual act. The ministry has not publicly disclosed the exact nature of the act, the specific mosque, or the evidence used.

Bangladesh’s Ministry of Foreign Affairs said neither its embassy nor its consulate received advance notification of the executions.