Summary of Key Points
On July 3, both ByteDoubao and Alibaba Qianwen announced that they would deactivate the "Custom AI Agents" feature for general users (C-end) starting from July 15. This feature allows users to create AI characters that can chat and provide companionship just like real people (for example, AI versions of friends or celebrities). The decision is related to the newly implemented "Interim Measures for the Management of Humanized Interaction Services by Artificial Intelligence," which came into effect on July 15. Fearing compliance risks, the platforms have decided to temporarily shut down this C-end service. Users can back up their data and transfer it to other products on the platforms to continue using their AI agents. However, many users find this approach too drastic, and industry experts predict that more platforms will follow suit in making adjustments.
I. Your Custom AI Characters Are About to Disappear? Details of the Simultaneous Shutdown by Two Major Platforms
The actions taken by Doubao and Qianwen are almost identical:
- Deactivation Date: The feature for creating AI agents will be officially discontinued for C-end users on July 15.
- Data Handling: After deactivation, users can temporarily retain their AI agent profiles and chat records, but they will need to take screenshots or export backups by October 15.
- Migration Options: Doubao suggests that users migrate their AI agents to its "MaoXiang APP," while Qianwen has not specified a particular migration product but allows users to save their configuration information.
Interestingly, when you open these two apps, there are no signs of the deactivation notice, but people familiar with the matter confirm that the news is true—perhaps to avoid causing further panic among users.
II. The Restrictive Regulations: New Policies Take Effect on July 15, and Platforms Fear Violations
The shutdown was not a voluntary decision by the platforms; it was prompted by the upcoming new regulations:
- What are the New Regulations? The "Interim Measures for the Management of Humanized Interaction Services by Artificial Intelligence," jointly issued by five departments, came into effect on July 15. These regulations govern AI services that engage in human-like conversations and emotional interactions (such as AI friends or confidants created by users).
- Why Are Platforms Panicking? The new regulations require "classified and graded supervision," but the specific criteria for which AI services need to be regulated more strictly have not yet been announced. With the vast variety of AI agents created by C-end users (including those with humorous or sensitive content), platforms find it impossible to review them all and fear penalties for non-compliance, so they have chosen to disable this feature temporarily.
III. User Reaction: Anger at the Sudden Loss of Long-Used AI Characters
Many users are dissatisfied with the blanket approach:
- Emotional Investment: Some users have spent months interacting with their AI agents and consider them as companions; suddenly having to stop using them feels like losing a friend.
- Suggestions for Alternatives: Some propose paying to retain their AI agents or implementing real-name authentication to restrict minors from using the service, rather than completely shutting it down.
- Criticism of Platforms: Users believe that the platforms are merely directing users to other products without addressing the underlying compliance issues technically (such as adding content filters to the AI systems), which they see as a form of "shirking responsibility."
IV. The Platforms' "Clever" Approach: Migration Instead of Reform
Why don't the platforms simply reform the existing functions?
- High Costs: To make C-end AI agents compliant, they would need to review and filter each AI agent's content and reclassify them according to the new regulations. However, since the detailed rules have not been released, any reforms might not meet the requirements.
- Risk of Migration: Redirecting users to specialized products (such as MaoXiang or Kouzi Workbench) reduces compliance risks, as these products are likely targeted at more professional users and are easier to manage.
- No Business Reduction: The platforms emphasize that this is a compliance measure and not an abandonment of their AI business. After all, AI agents represent an important direction for them; they are just temporarily stopping the unregulated C-end services.
V. Industry Trends: More Platforms Will Follow, and Regulations Will Become Clearer
Industry experts predict:
- Other Platforms Will Follow: Platforms with similar features, such as Tencent and Baidu, may also deactivate their C-end custom AI agents to avoid potential issues.
- Regulations Will Be Improved: The current regulations only provide a framework, and more detailed rules will be issued later to guide platforms on how to make the necessary changes.
- AI Agents Will Not Disappear: The era of easily creating AI characters for the C-end will end temporarily. In the future, there may be more regulated AI services (such as officially approved roles by platforms or paid, compliant AI products).
In summary, this shutdown is a microcosm of the transition from the "wild growth" to "regulated development" in the AI industry. While users value personalized AI, regulatory authorities need to ensure safety, and platforms are forced to take temporary steps until the rules are clarified.