# China's New AI Rules 2024: 4 Key Takeaways

> Canonical: https://dialzara.com/blog/chinas-new-ai-rules-2024-4-key-takeaways  
> Published: 2024-06-11  
> Summary: Key takeaways from China's new AI rules in 2024 covering content moderation, training data, user protection, and security assessment for AI services. Understand the regulations and business impacts.

_China's new AI rules create a two-tier system that could impact your business. Here's what you need to know to stay compliant and keep access._

## Key points

- Two-tier system: public AI faces strict rules, internal use stays flexible
- Budget for costs: security assessments and certifications now required
- Label all AI content clearly to meet transparency requirements
- Avoid penalties: violations range from warnings to full service shutdown

[China](https://en.wikipedia.org/wiki/China) has introduced new regulations for artificial intelligence ([AI](https://en.wikipedia.org/wiki/Artificial_intelligence)) services offered to the public within the country. These rules aim to ensure the responsible development and use of [AI](https://en.wikipedia.org/wiki/Artificial_intelligence) technologies while promoting innovation.

**Key Requirements:**

-   Content moderation: AI chatbots must not generate content that incites subversion, terrorism, extremism, or obscenity.
-   Training data: Strict guidelines for sourcing and using training data for AI models.
-   Content tagging: Generated content must be clearly labeled as AI-generated.
-   User protection: Measures to protect user privacy and prevent misuse of personal data.
-   Security assessment and filing: AI service providers must undergo security assessments and file with relevant authorities.

**What is Covered?**

The rules apply to generative AI services offered to the public in China, such as:

-   Chatbots
-   Virtual assistants
-   Content generation platforms

Exceptions include:

-   [Internal enterprise use of AI](https://dialzara.com/blog/ai-personal-assistant-for-business-task-automation/)
-   Research applications
-   AI services provided outside of China (unless accessible to the Chinese public)

**Securing Data and AI Models**

| Requirement | Details |
| --- | --- |
| Training Data Rules | No harmful or illegal data sources, clear data labeling guidelines, quality checks |
| Securing AI Models | Security assessments, model monitoring, handling user complaints |
| Content Tagging | Clearly label AI-generated content |
| Intellectual Property | Respect others' intellectual property rights |
| Personal Data | Process personal data with consent or legal basis |
| Fair Competition | Not use AI for monopolistic or unfair practices |

**Following the Rules**

| Compliance Step | Details |
| --- | --- |
| Security Assessments | For AI models with potential for widespread impact |
| Algorithm Filing | Provide details about algorithms to the government |
| Content Moderation | Monitor and moderate AI-generated content |
| Certifications | Obtain necessary certifications and licenses |
| Ethical Practices | Respect intellectual property rights and ethical business practices |
| Fair Competition | Avoid using data for monopolistic or unfair competition |
| User Protection | Ensure AI does not endanger health or infringe on rights |

Potential penalties for non-compliance include warnings, fines, service suspension, license revocation, and legal action.

**Business Impact**

-   Costs: Higher costs for businesses, especially SMEs, due to compliance measures like hiring staff, developing content moderation systems, obtaining certifications, and conducting security assessments.
-   Benefits: Increased user trust, enhanced reputation, improved risk management, and compliance with international standards.

| Aspect | Positive Impact | Negative Impact |
| --- | --- | --- |
| Compliance Costs |  | Higher costs for SMEs |
| User Trust | Increased trust and confidence |  |
| Market Position | Enhanced reputation and positioning |  |
| Risk Management | Improved risk management and mitigation |  |
| International Standards | Compliance with international standards and best practices |  |

[China's new AI rules](https://dialzara.com/blog/china-unveils-rules-for-generative-ai-services/) aim to promote [responsible AI development](https://dialzara.com/blog/10-top-tools-for-ethical-ai-development-2024/) while ensuring national security, ethical standards, and user trust. Businesses must understand and follow these rules to avoid penalties and maintain their reputation.

## Related video from YouTube

## Takeaway 1: Who and What is Covered

### What [AI](https://en.wikipedia.org/wiki/Artificial_intelligence) Services are Included?

[Image: AI]

The new rules apply to generative AI services offered to the public within China. These services use AI technology to generate content like text, images, audio, video, and more. The rules cover a wide range of industries and applications, such as:

-   Chatbots
-   Virtual assistants
-   Content generation platforms

### Any Exceptions?

The rules do not apply to:

-   Internal enterprise use of AI
-   Research applications

Companies using generative AI for internal purposes, like data analysis or research, do not need to comply. Additionally, AI services provided outside of China are exempt, unless accessible to the Chinese public.

| Covered by the Rules | Not Covered by the Rules |
| --- | --- |
| Generative AI services offered to the public in China | Internal enterprise use of AI |
| Chatbots, virtual assistants, content generation platforms | Research applications |
|  | AI services provided outside of China (unless accessible to the Chinese public) |

## Takeaway 2: Securing Data and AI Models

### Training Data Rules

AI service providers must follow strict rules for their training data:

-   **No harmful or illegal data sources**
-   **Clear data labeling guidelines** to ensure accuracy
-   **Quality checks** on labeled data and samples

Providers must also:

-   Respect intellectual property rights and business ethics
-   Not use data for monopolistic or unfair competition
-   Not endanger people's health or infringe on rights like privacy

### Securing AI Models

Providers must implement security measures for their AI models:

-   **Security assessments** and filing for models with "social mobilization ability"
-   **Monitoring and updating** models throughout their lifecycle
-   **Complaint handling** for user requests on personal data

They must also:

| Requirement | Details |
| --- | --- |
| Content Tagging | Clearly label AI-generated content |
| Intellectual Property | Respect others' intellectual property rights |
| Personal Data | Process personal data with consent or legal basis |
| Fair Competition | Not use AI for monopolistic or unfair practices |

### Comparing Security Measures

| Measure | Pros | Cons |
| --- | --- | --- |
| Security Assessments | Identify risks | Time and resource-intensive |
| Algorithm Filing | Transparency, accountability | May reveal intellectual property |
| Data Labeling | Improve data quality | Requires significant effort |
| Model Monitoring | Detect security incidents | May miss some threats |
| User Complaints | Build trust, user satisfaction | Increased support requests |

###### sbb-itb-ef0082b

## Takeaway 3: Following the Rules

### Steps to Comply

To follow China's new AI rules, providers must take these steps:

-   Assess the security of AI models with potential for widespread impact
-   File details about their algorithms with the government
-   Moderate and clearly label AI-generated content
-   Obtain necessary certifications and licenses
-   Respect intellectual property rights and ethical business practices
-   Avoid using data for monopolistic or unfair competition
-   Ensure AI does not endanger people's health or infringe on rights like privacy

### Consequences of Non-Compliance

If providers fail to comply, they may face:

-   Warnings and notifications from authorities
-   Fines and financial penalties
-   Suspension or termination of their services
-   Revocation of certifications and licenses
-   Legal action and potential criminal liability in severe cases

Regulatory bodies will monitor compliance and investigate reports of violations. Providers must cooperate with authorities and take corrective action to avoid penalties.

| Compliance Step | Details |
| --- | --- |
| Security Assessments | For AI models with potential for widespread impact |
| Algorithm Filing | Provide details about algorithms to the government |
| Content Moderation | Monitor and moderate AI-generated content |
| Certifications | Obtain necessary certifications and licenses |
| Ethical Practices | Respect intellectual property rights and ethical business practices |
| Fair Competition | Avoid using data for monopolistic or unfair competition |
| User Protection | Ensure AI does not endanger health or infringe on rights |

| Potential Penalty | Description |
| --- | --- |
| Warnings | Notifications from authorities about non-compliance |
| Fines | Financial penalties for violations |
| Service Suspension | Temporary or permanent suspension of services |
| License Revocation | Revocation of certifications and licenses |
| Legal Action | Potential legal action and criminal liability in severe cases |

## Takeaway 4: Business Impact

### Costs for Businesses

China's new AI rules may lead to higher costs for businesses, especially small and medium-sized enterprises (SMEs). These costs could include:

-   Hiring staff to manage AI-generated content and ensure algorithm transparency
-   Developing and implementing content moderation systems
-   Obtaining required certifications and licenses
-   Conducting regular security assessments and audits

While these costs may be a burden for SMEs, they are necessary to comply with the regulations and maintain user trust.

### Benefits of Following the Rules

Following China's new AI rules can bring several advantages to businesses, including:

| Benefit | Description |
| --- | --- |
| User Trust | Increased trust and confidence in AI-generated content |
| Market Position | Enhanced reputation and positioning |
| Risk Management | Improved risk management and mitigation |
| International Standards | Compliance with international standards and best practices |

By complying with the regulations, businesses can demonstrate their commitment to [ethical and responsible AI development](https://dialzara.com/blog/ethical-voice-ai-for-business-best-practices/) and deployment.

### Comparing Business Impacts

| Aspect | Positive Impact | Negative Impact |
| --- | --- | --- |
| Compliance Costs |  | Higher costs for SMEs |
| User Trust | Increased trust and confidence |  |
| Market Position | Enhanced reputation and positioning |  |
| Risk Management | Improved risk management and mitigation |  |
| International Standards | Compliance with international standards and best practices |  |

✓ indicates a positive impact, while ✗ indicates a negative impact.

By understanding the key points from China's AI regulations, businesses can adjust their strategies to comply, reduce risks, and take advantage of the opportunities presented by AI. As the AI landscape continues to change, it is crucial for companies to stay informed about ongoing developments in AI governance and adjust their approaches accordingly.

### Key Points

-   The rules cover generative AI services offered to the public in China, such as chatbots, virtual assistants, and content generation platforms.
-   Providers must follow strict guidelines for training data, including:
    -   No harmful or illegal data sources
    -   Clear data labeling guidelines
    -   Quality checks on labeled data and samples
-   Security measures for AI models include:
    
    | Measure | Details |
    | --- | --- |
    | Security Assessments | For models with potential for widespread impact |
    | Algorithm Filing | Provide details about algorithms to the government |
    | Content Moderation | Monitor and moderate AI-generated content |
    | User Complaints | Handle user requests on personal data |
    
-   Potential penalties for non-compliance:
    
    | Penalty | Description |
    | --- | --- |
    | Warnings | Notifications from authorities |
    | Fines | Financial penalties |
    | Service Suspension | Temporary or permanent suspension of services |
    | License Revocation | Revocation of certifications and licenses |
    | Legal Action | Potential legal action and criminal liability in severe cases |
    
-   Potential business impacts:
    
    | Positive | Negative |
    | --- | --- |
    | Increased user trust | Higher costs for SMEs |
    | Enhanced reputation and positioning |  |
    | Improved risk management |  |
    | Compliance with international standards |  |
    

## FAQs

### What is [China](https://en.wikipedia.org/wiki/China)'s new AI law?

[Image: China]

China has introduced new rules to regulate artificial intelligence (AI) technologies used in public services. The law covers AI services like chatbots, virtual assistants, and content generators offered to the public within China. It sets guidelines for:

-   Training data sources and quality
-   Security measures for AI models
-   Compliance requirements for providers

The goal is to ensure national security, ethical standards, and user trust while promoting AI innovation.

### Are there any exceptions?

Yes, the rules do not apply to:

-   AI research and development
-   AI services provided outside of China (unless accessible to the Chinese public)

However, providers must still follow relevant regulations and ensure their AI models do not pose risks to national security or public interests.

### How will businesses be impacted?

The new rules will require businesses to invest in compliance measures, such as:

| Compliance Measure | Description |
| --- | --- |
| Data Labeling | Clear guidelines for labeling training data |
| Security Assessments | Evaluating AI model security risks |
| Content Moderation | Monitoring and moderating AI-generated content |

While these measures may increase costs, they will also:

-   Enhance user trust and reputation
-   Improve risk management
-   Align with international standards

### Key Takeaways

-   China's new AI law regulates public-facing generative AI services like chatbots and content generators.
-   It sets rules for training data, AI model security, and compliance requirements.
-   The goal is to ensure national security, ethical standards, and user trust.
-   Businesses must invest in compliance measures like data labeling, security assessments, and content moderation.
-   Following the rules can enhance user trust, improve risk management, and align with international standards.

---

_By Dialzara Team._