AI+资本市场 (AI + Capital Markets): CSRC to Pilot Fintech Innovation Program

Key Points

AI+资本市场 (AI + Capital Markets) is getting a China-wide regulatory push from the China Securities Regulatory Commission (CSRC 证监会).

The announcement came from CSRC Vice Chairman Li Chao (Lǐ Chāo 李超) at the 2025 Financial Street Forum’s Fintech Conference on October 29.

The plan is a phased fintech pilot called “Artificial Intelligence + Capital Markets.”

It focuses on accelerating high‑value AI use cases, building shared computing and data infrastructure, and strengthening risk controls and oversight.

Quick take: why this matters to investors, founders, techies and marketers

The CSRC is signaling that AI in capital markets will move from isolated experiments to regulated, scalable deployments.

That means firms that treat AI as an isolated R&D toy will fall behind.

It also means new opportunities for platform providers, secure model vendors, data providers, and compliance tooling startups.

And it raises the bar on data governance, model auditing and human-in-the-loop controls for production AI in finance.

What the CSRC announced — headline program goals

The pilot pursues three clear priorities.

• Accelerate high‑value AI use cases in core capital‑market scenarios.
• Build shared compute, models and data infrastructure to reduce costs and improve interoperability.
• Harden risk controls and regulatory oversight across model lifecycles.

Who will be involved

The pilot will encourage participation from securities firms, fund managers and futures companies.

It will test production use cases under national and industry rules and only where risks are controllable.

Three core priorities of the pilot — in detail

1. Focus on high‑value application scenarios and industry integration

The CSRC plans to identify and prioritize business scenarios where AI can add clear value.

Examples called out include client service automation, investment research, risk management, and operations.

The pilot will encourage firms to scale AI‑driven service models and to document repeatable best practices.

Why this matters:

• Scalable playbooks: Documented implementations let other firms replicate what works.
• Commercial pressure: Firms that can show measurable process or cost improvements will get competitive advantage.

2. Strengthen foundational support: shared compute, models and data

The CSRC is proposing industry-level infrastructure to lower adoption costs and enable safer, interoperable AI systems.

Key proposals include:

• Accelerating construction of industry public intelligent computing infrastructure to reduce the compute cost burden on firms.
• Exploring an industry model platform and developing a matrix of multimodal, domain‑specific large models and intelligent agents to share across participants.
• Piloting “data elements × capital markets” (Shùjù yàosù × zīběn shìchǎng 数据要素×资本市场) initiatives to build high‑quality, industry‑specific datasets for model training and secure deployment.
• Building an industry shared knowledge base that aggregates critical domain knowledge across market businesses.
• Working toward an industry AI standards system to improve normative, compatible and secure application of AI in securities, fund and futures firms.

Practical impact:

• Lower cost to entry: Shared compute and shared models reduce duplication and free capital for product development.
• Interoperability: A model matrix and standards reduce integration friction between vendors and buyers.
• Data quality focus: Sector‑specific datasets make domain models more accurate and auditable.

3. Improve risk prevention and harden the security framework

The CSRC emphasized an end‑to‑end risk control system that covers model development, deployment and iteration.

Key risk measures listed are:

• Comprehensive model safety risk assessments, including model applicability and data‑use risks.
• Strengthening the “human‑in‑the‑loop” safeguard for critical decision points so that ultimate decisions remain with people and systemic risk is avoided.
• Enhancing data and network security controls to prevent leakage of sensitive information and protect AI systems from cyber threats.
• Establishing differentiated, scenario‑based regulatory approaches and a flexible institutional framework for AI applications.

Why this is important:

• Regulatory clarity: Firms that adopt strong model governance can reduce regulatory friction later.
• Operational safety: Human oversight prevents blind automation in sensitive market activities.
• Security posture: Secure pipelines and networks limit the risk of data exposure and adversarial attacks.

Context and industry uptake — what’s already happening

The CSRC has been promoting “AI+” and “data elements ×” pilots in recent years.

Securities firms, fund managers, and futures companies have already adopted AI tools in client services, research, risk control and operations.

That early adoption creates a favorable trend and a set of real business cases to scale.

The CSRC reiterated that while it supports innovation, applications must comply with national and industry requirements and put investor protection and data security first.

What market participants should expect next

The CSRC’s near‑term steps likely include selecting pilot participants, defining prioritized scenarios, and outlining technical and compliance standards for model and data governance.

Market participants should prepare by:

• Assessing where AI can generate measurable value in workflows and processes.
• Investing in secure, auditable model development lifecycles and robust human oversight processes.
• Engaging with industry initiatives for shared models, datasets and computing resources to avoid duplication and improve interoperability.

Actionable checklist for investors, founders and tech teams

• Investors: Prioritize startups with strong model governance, domain datasets and security-first architectures.
• Founders: Build product integrations that can plug into shared compute and model platforms.
• Engineering teams: Document model lineage, data provenance, and human‑in‑the‑loop checkpoints.
• Compliance teams: Start drafting scenario‑based regulatory mappings so pilots can scale into full production.

How this shifts the fintech landscape in China

The pilot represents a shift from point solutions to an industry‑level approach on AI infrastructure and governance.

It favors players who can deliver secure, auditable, and interoperable components — not just clever models.

For global investors and partners, the emphasis on shared resources and standards makes collaboration easier, provided partners meet the security and compliance requirements set by regulators.

Bottom line

The CSRC’s “AI+资本市场 (AI + Capital Markets)” pilot is a coordinated push toward scalable, secure AI in China’s capital markets.

Prepare for a landscape where shared compute, domain models, data governance and robust human oversight are table stakes.

AI+资本市场 (AI + Capital Markets)

References

• CSRC: Explore “AI+Capital Markets” fintech innovation pilot – Shanghai Securities News (上海证券报)
• China Securities Regulatory Commission (CSRC) official website – China Securities Regulatory Commission (证监会)
• 证监会：探索开展“AI+资本市场”金融科技创新专项试点 – Eastmoney (东方财富)