The inexorable rise of Artificial Intelligence technology and advanced data analytics across the financial sector is creating rapidly evolving regulatory and compliance challenges.

KPMG’s new report, AI Compliance in Control, examines the pressures institutions face and what action needs to be taken in an era when public policies regarding AI are developing more slowly than the technology.

With the proliferation of digital data for AI applications, the paper suggests establishing governance for data and algorithms amid heightened public awareness of data privacy and influence. Data used for system learning should be stored securely, with procedures in place to demonstrate compliance.

Along with other global fintech hubs, Hong Kong is looking to develop policies and regulations to promote the ethical and transparent adoption of the fast developing technology.

In particular, in May 2019 the Hong Kong Monetary Authority released a circular urging  Authorised Institutions to adopt the “Ethical Accountability Framework” by the Privacy Commissioner for Personal Data (PCPD) regarding the collection and use of personal data.

In common with its peers in Europe and the United States, Hong Kong must grapple with a host of regulatory and ethical challenges.

Key highlights:

• AI can help identify and mitigate risk sooner – but the opacity of some AI tools may make it more challenging to explain AI outcomes to regulators and consumers
• AI may allow creditors to more accurately model and price risk – but potential downsides include “unexplained” credit decisions, learned bias or other fair lending and consumer compliance risks
• AI can both be used to highlight cybersecurity vulnerabilities while also potentially introducing new cyber risks, so cybersecurity must be integral  
• The financial services industry, regulators and consumers – in Hong Kong and around the world – all have an interest in ensuring AI outcomes meet rigorous standards
• Code of conduct must be put in place; data analytics, audit trails and management reviews should evaluate AI outcomes

 

Download PDF (844 KB)