Hong Kong companies need to treat artificial intelligence governance as a board-level responsibility rather than an information technology project, with directors maintaining human accountability for AI-assisted decisions, Securities and Futures Commission (SFC) Chairman Kelvin Wong said.
Speaking at the Corporate Governance Conference 2026, Wong said companies needed to close a growing gap between the speed of AI adoption and the development of governance frameworks to oversee its use.
“The challenge is not merely how quickly companies adopt AI, but how effectively they turn artificial intelligence into accountable intelligence – and ultimately into performance that lasts,” Wong said in a keynote speech published by the SFC.
Wong identified governance, culture, and competence as three foundations for managing AI, arguing that technology can improve productivity while also increasing risks around opacity, third-party dependencies, cybersecurity, and accountability.
His comments come as AI becomes increasingly prominent in corporate disclosures in Hong Kong. A study by the Hong Kong Chartered Governance Institute and Wizpresso, based on annual and environmental, social and governance reports from more than 2,500 Hong Kong-listed companies, found that almost nine in 10 issuers referenced AI in their disclosures.
Fewer than 20% disclosed a structured AI governance framework covering oversight roles, policies and lifecycle controls.
The study found particularly weak disclosure on risk, compliance and incident management; AI governance and responsible use; executive ownership; and AI-related ESG issues.
It also found that larger issuers generally had more developed AI governance practices than smaller companies.
Wong said the governance challenge was particularly relevant to Hong Kong’s market structure.
The city has roughly 2,700 listed companies, but liquidity is highly concentrated, with about 9% accounting for more than 90% of average daily turnover, according to figures cited in his speech.
Among companies with market capitalizations below HK$4 billion, about 40% have a shareholder controlling more than half of the shares.
Such conditions can increase the importance of independent information and oversight, Wong said.
AI could help investors identify anomalies in corporate disclosures, but could also amplify selective narratives, misinformation and herd behavior in less liquid parts of the market.
He urged boards to establish clear responsibility for AI use and ensure they have enough information to oversee material deployments.
“Functions may be delegated; responsibility cannot,” Wong said, referring to directors’ duties under Section 465 of Hong Kong’s Companies Ordinance.
He said all directors should have basic AI literacy, while directors with technology, data, cybersecurity, or AI responsibilities would face higher expectations based on their relevant expertise.
Boards should also understand the AI use cases within their companies, assign named owners and ensure that controls match statements made to investors, he added.
The SFC’s own regulatory work has highlighted the cybersecurity dimension.
In June, the regulator told licensed corporations, licensed virtual asset service providers, and associated entities to strengthen defenses against AI-enabled cyberattacks, including through updated technology inventories, vulnerability management, access controls, third-party risk management, and incident response.
The SFC said frontier AI models could reduce the expertise, cost and time required to identify and exploit software vulnerabilities, while AI-enabled tools could facilitate phishing, social engineering, deepfake impersonation and reconnaissance.
Wong said companies should therefore develop firm-wide, board-endorsed AI guidelines covering strategy, capital allocation, risk and competence, while defining areas where human judgment should not be replaced.
He also called for greater discipline in AI investment. Boards should assess what business problem an investment addresses, expected returns and risks, as well as conditions that would prompt a company to expand, change or terminate the investment.
“Activity is not achievement. Expenditure is not value creation. And adoption is not transformation,” Wong said.
He added that AI should remain a tool for augmenting human capability rather than replacing judgment, complex risk assessment or ethical decision-making.
“AI may be artificial. Accountability must remain human. And trust must remain earned,” he said.