/ AI & the Agentic Economy

Legal certainty for systems that act on their own.

AI no longer just informs decisions — it acts on them. When software places orders, accepts terms and moves value autonomously, questions of authority, liability and compliance become legal questions. We help founders, financial institutions and corporates deploy AI in Hong Kong with risk allocated before an incident, not after.

Book a consultation →Explore our capabilities

/ Why this matters now

The law was written for humans who decide. Agentic systems challenge that assumption.

Hong Kong does not yet have a single dedicated AI statute. Instead, liability, contracting and licensing issues are addressed through a patchwork of existing laws and regulatory expectations, including the SFO, PDPO, HKMA guidance, SFC requirements and the VASP regime. These frameworks were not designed for software that acts with limited human intervention. Businesses that structure for this early will be better placed when the rules harden.

Authority

Did the AI have authority to bind you? When an agent places an order, accepts terms or executes an instruction, enforceability turns on agency, delegation and system design.

Liability

Developer, deployer or operator — when an autonomous system causes loss, responsibility should be allocated by design, not discovered in litigation.

Compliance

AI in trading, custody, payments or tokenised assets can trigger regulatory obligations that existing controls were never built to satisfy.

/ Who we advise

For the people deploying AI, not just studying it.

AI and agent startups

Liability structuring, IP ownership and contracting frameworks for autonomous products.

Financial institutions

AI in trading, compliance and operations — navigating SFC and HKMA supervisory expectations.

Asset managers using AI

Portfolio-management AI, SFC licensing, governance and model-risk design.

Web3 and DeFi projects

Agents interacting with smart contracts, tokenised assets and on-chain payment rails.

Corporates and boards

Governance and legal risk for agentic systems that make decisions or enter obligations.

Founders and GCs

/ How we help

Four problems the agentic economy creates — and how we help solve them.

AI agent liability and risk allocation

When an autonomous system causes harm or creates an obligation, responsibility is rarely obvious. We help allocate liability between developers, deployers and principals, and draft agreements that reflect how autonomous decision-making works in practice.

Autonomous transactions and enforceability

Agents can now place orders, execute trades and accept terms with limited oversight. We advise on whether those actions bind the principal, how authority and delegation should be structured, and how to build legal certainty into automated workflows.

AML/CFT and autonomous financial flows

Systems that touch financial flows, payment infrastructure or virtual assets can attract licensing, monitoring and reporting obligations. We map how the AMLO and VASP regime may apply and design controls before deployment.

AI and tokenised real-world assets

Using AI to manage or move tokenised real-world assets raises questions of custody, authority, classification and control. We structure tokenised-asset programmes, clarify regulatory treatment and document ownership and governance arrangements.

/ The regulatory map

No single AI law — but several regimes need to be read together.

We translate Hong Kong’s regulatory patchwork into a governance framework that can evolve with the rules.

SFO — AI in financial servicesPDPO — data & automated decisionsHKMA — AI in bankingSFC — algo & AI guidanceAMLO — autonomous flowsVASP — tokenised & on-chain

/ FAQ

Questions clients ask before they deploy.

Who is liable when an AI agent causes harm in Hong Kong?+

There is no dedicated AI-liability statute in Hong Kong. Responsibility is assessed by applying contract, tort, agency and regulatory principles to the facts. Depending on the structure, exposure may sit with the developer, deployer, operator or principal. We help allocate that risk before an incident occurs.

Is there a specific AI law in Hong Kong?+

Not yet. AI is currently governed through existing laws and regulatory guidance, including data protection law, financial-services regulation and sector-specific supervisory expectations. We build frameworks that are practical now and capable of adapting as the rules develop.

What SFC or HKMA requirements apply to AI in financial services?+

The recurring themes are governance, explainability, human oversight, model-risk management, testing and monitoring. For licensed firms and banks, the key is not just whether AI is used, but whether the firm can evidence appropriate controls around its design, deployment and review.

Who owns the IP in AI-generated content or code?+

Ownership depends on the facts, including the level of human input, the terms of the AI tool, the development arrangements and any assignment clauses. We structure development agreements, employment terms and IP clauses to support the intended ownership position.