Financial firms rank data quality as top AI compliance risk

Generative technology attracted interest from 77% of respondents.

Data quality is the biggest concern for financial institutions looking to use artificial intelligence (AI) in compliance.

The survey, conducted by Risk.net and Fenergo, gave data quality an aggregate concern score of 275, ahead of integration with existing systems at 236. 

Regulatory and compliance concerns scored 209, followed by implementation costs at 207.

Trust and reliability scored 172, whilst data privacy concerns reached 162. A lack of expertise scored 132 and technical issues 100. 

Change management and C-suite or board-level support ranked lower, with scores of 42 and 35 respectively.

Despite these concerns, financial institutions are considering a range of AI technologies. 

Generative AI was the most commonly cited, with 77% of respondents saying their organisation was considering its adoption. Machine learning followed at 68%.

Robotic process automation was cited by 47%, natural language processing by 46% and agentic AI by 44%. 

Another 40% said their organisations were considering predictive analytics, whilst 2% said they were considering no forms of AI.

Amongst respondents considering agentic AI, 66% said transaction monitoring was a priority use case. 

Fraud detection was cited by 55% and sanctions screening by 46%. KYC maintenance and customer onboarding were identified by 42% and 41%, respectively.

The report said legacy systems, fragmented data and manual processes were making AI implementation more difficult. It also identified governance, data management and shortages of staff with the required technical skills as challenges for compliance teams.

The survey gathered insights of 110 practitioners at banks and asset managers in Singapore, Malaysia and Australia.