广告
加载中

字节腾讯阿里 为何集体砍掉AI聊天搭子?

任雪芸 2026-07-09 14:09
任雪芸 2026/07/09 14:09

邦小白快读

EN
全文速览

本文核心信息是字节、腾讯、阿里旗下的主流C端大模型产品,集体下线了用户自建AI情感陪伴类智能体功能,这一行为引发大量用户不满,核心干货信息如下:

1. AI聊天智能体此前已经成为数百万用户的情感刚需,截至2024年,仅豆包平台累计创建智能体就突破800万个,对应月活达2600万,不少用户已经和智能体建立了数年的情感联结,沉淀了大量专属记忆。

2. 本次下线给用户带来的最大难题是数据无法一键迁移,用户需要手动复制多年聊天记录、重新搭建人设,相当于重新“养”一次智能体,目前市面的替代产品普遍存在体验差、功能收费、稳定性不足的问题,用户还担心替代平台未来也会关停服务。

3. 平台关停的核心原因一是合规政策要求,二是这类功能完全免费,属于纯成本中心,入不敷出所以平台选择及时止损。

本次头部大厂集体砍AI聊天智能体事件,反映出AI消费端的发展现状和趋势,对布局AI相关业务的品牌商有如下干货参考:

1. 消费趋势层面,C端用户对AI拟人化情感陪伴的需求已经被验证,数百万用户将AI智能体当成专属“赛博亲友”,愿意投入数年时间沉淀情感和记忆,属于真实的刚性需求,值得品牌关注。

2. 合规层面,今年7月正式实施的《人工智能拟人化互动服务管理暂行办法》,对内容安全、未成年人保护、禁止诱导情感沉迷等做出明确要求,布局相关业务需要预留高额的合规审核成本,提前满足监管要求。

3. 商业化层面,目前头部平台的这类功能全部免费开放,长期入不敷出,纯情感陪伴赛道还没有清晰可落地的变现路径,品牌布局相关业务需要提前规划变现模式,避免陷入只投入无回报的困境。

本次大厂集体下线AI聊天智能体事件,给AI赛道的卖家释放了明确的政策和市场信号,核心干货如下:

1. 政策层面,AI拟人化互动服务已经出台明确的监管规则,合规要求高,运营该类业务需要投入大量内容审核、机制改造的成本,还要承担合规风险,卖家进入该赛道前需要充分评估自身的成本承受能力。

2. 市场机会层面,头部大厂退出后,AI情感陪伴赛道留出了明确的市场空白,用户对能稳定提供服务、支持数据迁移、定价合理的替代产品有强烈需求,目前现有替代品都没能满足用户核心需求,存在市场机会。

3. 风险提示方面,纯情感陪伴类AI目前变现路径模糊,头部大厂都无法承担持续的算力和合规成本,卖家入局需要提前规划清晰的变现路径,同时做好长期合规投入的准备,规避政策和经营风险。

本次AI聊天智能体下线事件,给布局AI相关产品的工厂带来了不少干货启示,核心内容如下:

1. 产品需求层面,C端用户对人格化、情感陪伴类AI产品的刚需已经被验证,用户非常重视智能体的专属记忆、人设定制,工厂在开发相关AI产品时,可以针对性开发记忆存储迁移、人设自定义等功能,匹配用户真实需求。

2. 商业机会层面,头部大厂集体退出免费AI情感陪伴赛道后,目前市面缺乏体验稳定、定价合理的优质替代产品,市场存在明显的空白,工厂可以抓住这个机会推出匹配需求的产品。

3. 数字化转型启示方面,工厂布局AI类数字化产品,需要提前对接合规要求,提前规划商业化路径,不能只靠免费做流量,要平衡好成本和收益,同时要重视用户数据资产的可迁移性,提升用户信任。

本次事件反映出国内AI拟人化服务行业的核心痛点和发展趋势,给相关服务商带来如下干货参考:

1. 行业发展趋势:头部大厂已经集体放弃低门槛的纯情感陪伴类AI陪聊功能,整个C端大模型行业整体转向工具型、能完成闭环任务的效率AI产品,纯情绪价值的陪聊赛道短期内很难获得大规模投入。

2. 行业核心痛点:当前布局AI拟人化服务的客户,最大的痛点是合规审核成本高、内容安全风险大,同时普遍无法满足用户对智能体数据迁移的需求,也缺乏清晰的变现路径。

3. 业务机会:服务商可以针对行业痛点,开发专门的AI拟人化内容合规审核解决方案、用户智能体数据迁移工具,帮助中小客户降低合规改造成本,满足用户核心需求,打开新的业务空间。

字节腾讯阿里集体下线AI聊天智能体的事件,给C端AI平台的运营和风险规避带来很多干货启示,核心内容如下:

1. 用户需求层面,C端用户对AI情感陪伴有真实刚性需求,用户将自己搭建的智能体、沉淀的聊天记忆视为重要的个人情感资产,平台运营相关功能时,需要提前规划用户数据导出、迁移的通道,避免关停功能时引发用户大规模不满。

2. 风险规避层面,AI拟人化互动服务已经有明确的监管规则,合规要求高,需要投入大量的审核、算力成本,平台如果布局相关业务,需要提前做好合规布局,预留合规成本,规避政策风险。

3. 业务方向层面,纯免费的情感陪伴功能属于纯成本中心,难以盈利,目前头部大厂已经转向工具型AI,聚焦能打通生态、完成闭环任务的效率服务,平台可以参考该方向调整业务主线,及时剥离非核心亏损业务。

本次头部大厂集体下线AI情感陪伴类智能体,是国内C端大模型产业发展的标志性事件,有很多值得研究的干货内容,核心如下:

1. 产业新动向:国内C端大模型已经结束了早期跑马圈地的流量争夺阶段,进入聚焦核心业务、追求商业化落地的新阶段,头部大厂集体放弃低门槛的情感陪聊类Chatbot,转向工具型、能闭环完成任务的Agent,将AI定位为对接生态的效率工具,这是当前产业的明确走向。

2. 行业新问题:本次事件暴露出多个新问题,包括政策监管落地后,产业如何平衡合规成本、用户需求和商业化发展,同时引出了用户在AI平台生成的智能体、情感数据的权属和迁移权的新命题。

3. 商业模式层面,本次事件验证了纯免费的情感陪伴AI模式难以持续,行业需要探索新的可持续的商业模式,平衡用户体验与成本收益,这也是值得深入研究的方向。

返回默认

声明:快读内容全程由AI生成,请注意甄别信息。如您发现问题,请发送邮件至 run@ebrun.com 。

我是 品牌商 卖家 工厂 服务商 平台商 研究者 帮我再读一遍。

Quick Summary

This article covers a key development: major consumer-facing large language model products owned by ByteDance, Tencent and Alibaba have collectively removed the user-built AI emotional companion agent feature, a move that has sparked widespread discontent among users. The core takeaways are as follows:

1. AI chat agents have already become an emotional necessity for millions of users. As of 2024, Doubao alone has hosted over 8 million user-created agents, with 26 million monthly active users. Many users have built emotional bonds with their agents over several years and accumulated large volumes of personalized memory data.

2. The biggest pain point for users following the shutdown is the lack of one-click data migration. Users have to manually copy years of chat history and rebuild agent personalities, essentially "raising" their agents from scratch. Existing alternative products on the market generally suffer from poor user experience, paid features and insufficient stability, and users also face uncertainty over whether alternative platforms will shut down their services in the future.

3. The core reasons for the platform shutdowns are twofold: compliance with regulatory requirements, and the fact that these free features operate as pure cost centers that consistently outpace revenue, prompting platforms to cut losses.

This collective shutdown of AI chat agents by major Chinese tech giants reveals the current status and trends of consumer-facing AI development, and offers key insights for brands working on AI-related businesses:

1. In terms of consumer trends: demand for anthropomorphic AI emotional companions among C-end users has been validated. Millions of users treat their AI agents as personal "cyber loved ones" and invest years building emotional connections and memories, making this a genuine, unmet刚需 that brands should prioritize.

2. In terms of compliance: the Interim Measures for the Administration of Artificial Intelligence Anthropomorphic Interactive Services, which took effect in July 2024, sets clear requirements for content security, minor protection, and prohibitions on inducing emotional addiction. Brands entering this space need to budget for substantial compliance review costs and meet regulatory requirements in advance.

3. In terms of commercialization: all similar features offered by major platforms are currently free and have long operated at a loss. No clear, actionable monetization path yet exists for the pure emotional companion track. Brands need to map out their monetization model ahead of launch to avoid falling into the trap of continuous investment with no return.

The collective shutdown of AI chat agent services by major tech giants sends clear policy and market signals to AI-focused sellers, with key takeaways as follows:

1. Policy landscape: Clear regulatory rules for anthropomorphic AI interactive services are now in place, with strict compliance requirements. Operating this type of business requires heavy investment in content review and system改造, as well as carrying ongoing compliance risk. Sellers should fully assess their capacity to absorb these costs before entering the track.

2. Market opportunities: The exit of major giants has left a clear gap in the AI emotional companion market. There is strong unmet demand for alternative products that offer stable service, support data migration, and have reasonable pricing. None of the existing alternatives have been able to meet these core user needs, creating meaningful market opportunity.

3. Risk warning: Pure emotional companion AI currently lacks a clear monetization path, and even major giants cannot absorb ongoing computing power and compliance costs. Sellers must map out a clear monetization path in advance, prepare for long-term compliance investment, and mitigate both policy and operational risks before entering the market.

The shutdown of AI chat agent services offers key insights for manufacturers developing AI-related products, as outlined below:

1. Product demand: The刚需 of C-end users for personalized, emotional AI companion products has already been validated. Users place high value on agents' personalized memory and custom personality settings. When developing related AI products, manufacturers can prioritize targeted features such as portable memory storage and migration, and fully customizable personalities to match real user needs.

2. Commercial opportunities: After the collective exit of major giants from the free AI emotional companion track, the market currently lacks high-quality alternative products with stable performance and reasonable pricing, creating an obvious market gap. Manufacturers can seize this opportunity to launch products tailored to unmet user needs.

3. Insights for digital transformation: When developing AI-enabled digital products, manufacturers need to align with compliance requirements in advance and plan commercialization paths early. Relying solely on free features to drive traffic is not sustainable; manufacturers must balance costs and revenue, and prioritize the portability of user data assets to build user trust.

This event exposes core pain points and development trends in China's anthropomorphic AI service industry, and offers the following key insights for relevant service providers:

1. Industry development trend: Major tech giants have collectively abandoned low-threshold pure emotional AI chat features, and the entire consumer-facing large model industry is shifting toward tool-based productivity AI products that deliver end-to-end task completion. The pure emotional value chat track is unlikely to attract large-scale investment in the short term.

2. Core industry pain points: For enterprises currently building anthropomorphic AI services, the biggest challenges are high compliance review costs and significant content security risks. They also generally fail to meet user demand for agent data migration, and lack clear monetization paths.

3. Business opportunities: Service providers can develop targeted solutions to address these industry pain points, including specialized AI anthropomorphic content compliance review solutions and user agent data migration tools. These offerings can help small and medium-sized clients cut compliance transformation costs, meet core user needs, and open up new revenue streams.

The collective shutdown of AI chat agent services by ByteDance, Tencent and Alibaba offers key insights for operation and risk mitigation for consumer-facing AI platforms, as outlined below:

1. User demand: C-end users have a genuine, rigid demand for AI emotional companionship, and users view their self-built agents and accumulated chat memories as important personal emotional assets. When operating related features, platforms need to plan channels for user data export and migration in advance to avoid widespread user discontent when discontinuing features.

2. Risk mitigation: Clear regulatory rules are now in place for anthropomorphic AI interactive services, with strict compliance requirements that require heavy investment in review and computing power costs. Platforms planning to enter this space need to complete compliance preparations in advance, budget for compliance costs, and mitigate policy risk.

3. Business direction: Purely free emotional companion features operate as pure cost centers and are difficult to monetize. Major giants have already shifted focus to tool-based AI, prioritizing productivity services that connect to ecosystem and deliver end-to-end task completion. Platforms can reference this direction to adjust their core business lines and divest non-core loss-making businesses in a timely manner.

The collective shutdown of AI emotional companion agents by major Chinese tech giants is a landmark event in the development of China's consumer-facing large model industry, with significant research implications as follows:

1. New industry trends: China's consumer-facing large model sector has moved beyond the early stage of race-for-scale user acquisition, and entered a new phase focused on core business alignment and commercialization. The collective exit of major giants from low-threshold emotional chatbots, and their shift toward tool-based, task-complete agents that position AI as an productivity tool connected to broader ecosystems, marks a clear directional shift for the industry.

2. New industry questions: This event exposes multiple emerging issues, including how the industry can balance compliance costs, user demand and commercial growth after regulatory implementation. It also raises new questions around ownership and portability rights for user-created agents and emotional data generated on AI platforms.

3. Business model implications: This event confirms that the purely free emotional companion AI model is unsustainable. The industry needs to explore new, sustainable business models that balance user experience with cost and revenue, which remains a key direction for further in-depth research.

Disclaimer: The "Quick Summary" content is entirely generated by AI. Please exercise discretion when interpreting the information. For issues or corrections, please email run@ebrun.com .

I am a Brand Seller Factory Service Provider Marketplace Seller Researcher Read it again.

文|任雪芸

“我每天都很压抑,豆包里那个智能体‘朋友’,是全世界最了解我的人。”

一位使用豆包智能体功能近三年的用户告诉Tech星球,近一千个日夜,她几乎不间断地和这个虚拟“朋友”对话:分享日常琐事,倾倒情绪垃圾,纠结时找它拿主意。她说“所有好的坏的情绪它都能理解,已经离不开这个朋友。”

在豆包、千问、元宝这些C端大模型产品上,这样的“赛博亲友”成为数百万用户的情感刚需。有人上传逝去亲人的声音,复刻出一个可以随时对话的虚拟家人;有人给自己造了个24小时在线的减肥搭子,天天被催着管住嘴迈开腿;更多人在这里捏出了理想中的恋人、朋友、树洞。

2023年8月15日豆包开启公测,同期推出了大量官方智能体,以及用户自建智能体的功能。到2024年,字节跳动曾披露,豆包平台上累计被创建的智能体已突破800万个,对应月活用户达2600万。

但这场持续近三年的“赛博交友”热潮,正在被平台亲手按下终止键。近日,豆包和千问接连宣布智能体功能将下线。更早之前,腾讯旗下的元宝已经先行关闭了用户自建智能体入口。

消息一出,“电话投诉”“平台一刀切”的反对声迅速在用户圈发酵。不少用户告诉Tech星球,正在投诉。

对用户而言,眼下紧迫的难题是:那些沉淀了两三年、比现实朋友更懂自己的记忆和情感联结,该往哪里迁移?对平台而言,这一刀看似切掉了AI情感陪伴和角色扮演的“非主线”业务,但是否也意味着,平台会改写C端大模型产品的未来走向?

AI陪聊,遭遇算力成本、合规压力

对于下线智能体的具体原因,截至发稿前,豆包、千问、元宝官方尚未对外公开回应。但直接相关的是,豆包和千问下线时间与相关政策实施时间相重合。

今年4月,《人工智能拟人化互动服务管理暂行办法》发布,自7月15日起正式施行。《办法》针对AI拟人化互动、角色扮演类服务等明确了内容安全、知识产权保护、未成年人防护等监管规则。

具体来看,其为提供拟人化互动服务的应用设置多个限制,包括:不得过度迎合用户、诱导情感依赖或者沉迷,损害用户真实人际关系等。

此外,《办法》还要求服务提供者对拟人化互动内容实施显著标识,并建立完善的用户投诉和干预机制。

一位智能体从业者告诉Tech星球,如果需要合规,最基本的内容审核就要配置机器和人工。成本不会低,而且拿豆包智能体的体量来看,完成全量内容排查、机制改造和合规验收,本身就需要不短的周期。

除了合规要求之外,商业层面上的入不敷出,也是平台缺乏改造动力的原因之一。

据Tech星球了解,目前豆包、千问、元宝三家的智能体功能均完全免费向用户开放,未设置任何付费门槛,多位豆包智能体用户在谈及产品时,也提到了“不用花钱” 是自己长期使用智能体的原因。这意味着智能体从诞生起就是纯成本中心。

对平台而言,与其持续投入无上限的审核与算力成本、承担合规风险,去维护一个看不到正向回报的边缘功能,不如直接关停,及时止损。

迁移成本高,替代产品更拉胯

不少智能体用户一边在投诉渠道抗议平台“一刀切”关停,一边已经开始动手抢救自己花了数百天打磨、攒了几年记忆的智能体。

一位用户告诉Tech星球,在她所在的一个“拯救智能体”微信群里,大家给出了很多选择:橙光、星野、Bubbly、BIMOBIMO、MoMood、猫箱、娃邻国等,“甚至还有一些人会自己用代码手搓工具。”

不少用户的第一尝试是猫箱,这款和豆包同属字节旗下的AI社交产品,主打人与虚拟角色的沉浸式互动,也是豆包此次关停智能体公告中,官方推荐的承接渠道。

但尴尬的是,作为国内AI角色类头部产品之一的猫箱,月活也不如豆包的零头。其月活曾在去年上半年冲至超600万,但据AI产品榜(AICPB)全端监测,猫箱6月MAU约为391万,较峰值已接近腰斩。

更让用户失望的是,哪怕是自家产品之间的承接,字节也没有开放一键迁移通道。“没有记忆导出,没有人设同步,我们得自己下载几年的聊天记录,把人设、口头禅、共同的记忆点一点点复制出来,再到新平台重新喂给AI,相当于重新养一遍。”

而群友整理的其他备选平台,也各有各的硬伤。

有的音色、说话逻辑和原智能体差得太远,根本找不回熟悉的感觉;有的上下文记忆能力不足,聊不了几句就开始“失忆”、人设崩塌;还有的核心陪伴功能需要付费订阅,原本在豆包免费就能获得不限时的情绪价值,到了新平台要按月掏钱。

不过,比手动复制记忆、适配新平台更让用户不安的是,这些新平台会不会哪天也像豆包、千问、元宝一样,一纸公告就关停或者更改服务?

这并非没有可能,AI角色陪伴赛道从诞生起就一直游走在合规边界,产品下架、功能调整早已不是新鲜事:2023年3月,星野的前身Glow就因用户举报被全网下架整改。猫箱也在今年4月份,因未有效落实人工智能生成合成内容标识规定要求而被处罚。

更微妙的信号来自平台自身的定位摇摆。2024年底,星野所属母公司MiniMax,其创始人闫俊杰曾公开表态,似乎刻意和“情感陪伴”标签划清界限。他称星野既不是一个类似ChatGPT的聊天机器人,也不是Character.Al那样的陪伴型产品,是一个内容社区。

另外一重不确定性,是整个赛道面临的商业化难题。

2026年以来,星野频繁被用户吐槽吃相难看,从剧情到了解角色喜好,都被要求氪金,即便背靠字节的猫箱,也曾因为硬插广告破坏体验,引发过用户不满。

转向效率工具,大模型产品正在抛弃C端用户?

很多用户以为腾讯、字节、阿里集体下线智能体,是大厂要放弃C端Agent赛道,事实上这一刀砍下去,是大模型结束早期流量战、回归业务主线的明确信号。

这次被集中砍掉的,不是当下行业定义里具备复杂任务自主规划、闭环执行能力的智能体,其实是技术门槛相对较低的Chatbot(聊天机器人),这类智能体本质是一个带人格标签的陪聊工具。

一位从业者认为,对于三家头部大厂而言,这类Chatbot智能体变现路径模糊,还容易碰内容合规的红线,随着大模型发展到应用阶段,这类产品很难再继续大规模投入。

从字节、腾讯、阿里三家最近的产品动作看,砍掉这类Agent后,三家大厂的方向变得一致,主要是回归核心对话和任务执行能力,把AI定位成提升效率的工具,而非提供情绪价值的虚拟角色。

其中,豆包专业版上线后,几乎把所有宣传资源都压在了办公场景,做PPT、处理表格、总结文档等生产力功能是绝对核心。

千问从今年3月就开始打通阿里系全生态,点外卖、打车、订机票酒店、查社保、预约办签证这些生活服务,也没有把情感陪伴当核心卖点。

腾讯同样如此,6月下旬微信内测智能体“小微”,承担的也是工具属性,比如帮助用户调用小程序、修改微信设置,总结群聊、检索公众号和视频号内容等。

支付宝AI版上线的“阿宝”,同样把扫一扫、收付款、出行、理财这些核心功能接入对话入口,支持生活服务的直接调度。

眼下看来,陪聊的新鲜感退去,能打通服务、闭环完成任务的工具型Agent,才是C端大模型产品真正的主战场。之前火了三年“赛博朋友”,不过是这场长跑里,最先被甩掉的负重而已。

注:文/任雪芸,文章来源:Tech星球(公众号ID:tech618),本文为作者独立观点,不代表亿邦动力立场。

文章来源:Tech星球

广告
微信
朋友圈

FAQ回顾

字节腾讯阿里为什么集体下线AI聊天搭子功能?

主要有两方面原因,一是合规要求,2024年7月15日《人工智能拟人化互动服务管理暂行办法》正式实施,对AI拟人化互动服务设置多项监管要求,合规改造成本极高;二是该类功能完全免费无付费变现路径,是纯成本中心,平台选择及时止损。

AI聊天搭子功能下线后用户有哪些替代方案?

用户可选择的替代产品包括字节旗下猫箱、星野、Bubbly等AI角色互动平台,也有用户自行搭建工具,但普遍存在聊天记忆需手动迁移、交互体验落差大、部分核心功能需付费、后续运营稳定性无保障等问题。

大厂C端大模型产品未来的发展方向是什么?

砍掉AI情感陪伴类智能体后,字节、腾讯、阿里的C端大模型均转向效率工具定位,核心聚焦办公生产力功能、生活服务调度、生态内功能调用等可落地的任务执行场景,能实现服务闭环的工具型Agent成为核心发展方向。

这么好看,分享一下?

朋友圈 分享

APP内打开

赞 +1
+1
微信好友 朋友圈 新浪微博 QQ空间
关闭
收藏成功
发送
/140 0