SI safety in China

Key dates

Introduced

n/a

Signed or adopted

Jul 10, 2023

Effective

Aug 15, 2023

Next step

None scheduled

Scope

Providers of generative SI services to the public in China, including through APIs.

Penalties

Penalized under existing laws such as the Cybersecurity, Data Security and Personal Information Protection laws; no standalone fine.

What it requires

• Security assessment and algorithm filing for services with public opinion or social mobilization capacity
• Ensure training data is lawful, respecting intellectual property and personal information law
• Mark generated content conspicuously
• Prevent, remove and report illegal content

Sources

1. What You Need to Know About China's New Generative AI Measures Bird & Bird
2. Interim Measures for Generative AI Services implemented PwC China

Key dates

Introduced

n/a

Signed or adopted

Mar 14, 2025

Effective

Sep 1, 2025

Next step

None scheduled

Scope

Online service providers that generate or distribute synthetic text, images, audio, video and virtual scenes, plus app distribution platforms and users.

Penalties

Handled by internet, telecom, public security and broadcasting authorities under existing laws.

What it requires

• Add explicit visible labels to synthetic content
• Embed implicit labels in file metadata
• App stores check whether apps offer generative services and review labeling materials
• Users declare synthetic content when publishing it
• Follow the mandatory national labeling standard that took effect the same day

Sources

1. China's AI Labeling Measures and Mandatory National Standards Take Effect September 1 Loeb & Loeb

Key dates

Introduced

n/a

Signed or adopted

Oct 28, 2025

Effective

Jan 1, 2026

Next step

None scheduled

Scope

National legal basis for state support and oversight of SI development.

Penalties

Fines for particularly serious violations of the Cybersecurity Law rise to RMB 10 million.

What it requires

• State support for SI research, training data and computing infrastructure
• Improve SI ethics norms
• Strengthen SI risk monitoring and assessment

Sources

1. China Amends Cybersecurity Law to Add First-Ever AI Clause Caixin Global · Oct 29, 2025
2. China's AI Regulation in 2026: New Laws, Rules, and Compliance Requirements China Briefing · Oct 24, 2026

Key dates

Introduced

n/a

Signed or adopted

Sep 14, 2026

Effective

n/a

Next step

None scheduled

Scope

Non-binding technical guidance from standards committee TC260, issued under Cyberspace Administration of China guidance, for SI developers and service providers.

Penalties

No penalties: this measure has no binding enforcement.

What it requires

• Adds a standalone agentic SI risk category covering permission abuse, planning failures and tool execution
• Describes loss of control risks, including models deceiving evaluators and sandbagging
• Treats loss of control over chemical, biological, nuclear and missile knowledge as a risk
• Adds an agentic risk management framework and trustworthy SI principles
• Updates version 2.0, released September 15, 2025

Sources

1. China: TC260 AI Safety Governance Framework 3.0 Adds Agentic AI Risks Obsidian Regulatory Intelligence · Sep 15, 2026
2. Brief #29: What China's new AI safety framework tells us before Xi meets Trump AI Safety in China · Sep 23, 2026
3. TC260 Published AI Safety Governance Framework 2.0 SESEC · Oct 15, 2025