Financial institutions across Asia-Pacific are racing to adopt artificial intelligence for compliance work, but they're discovering a fundamental problem: their data isn't ready for it. A new survey of banking and asset management professionals reveals that data quality has emerged as the single biggest concern for firms looking to deploy AI in compliance operations, far outpacing regulatory worries or technical hurdles.

What's Really Holding Back AI in Finance?

The survey, conducted by Risk.net and Fenergo and involving 110 practitioners at banks and asset managers in Singapore, Malaysia, and Australia, painted a stark picture of the obstacles ahead. Data quality received an aggregate concern score of 275, nearly 40 points higher than the second-place concern: integration with existing systems at 236. This gap is significant because it suggests that financial institutions understand the technical challenges of AI implementation, but they're most worried about the foundation everything else rests on.

The remaining concerns tell an interesting story about where the real pain points lie:

• Regulatory and Compliance Concerns: Scored 209, reflecting uncertainty about how regulators will evaluate AI systems in financial services.
• Implementation Costs: Reached 207, indicating budget constraints as firms weigh expensive infrastructure upgrades.
• Trust and Reliability: Scored 172, showing concerns about whether AI systems will make consistent, defensible decisions.
• Data Privacy: Reached 162, reflecting heightened awareness of customer data protection requirements.
• Lack of Expertise: Scored 132, pointing to a shortage of staff who understand both compliance and AI.
• Technical Issues: Scored 100, the lowest among major concerns, suggesting firms are less worried about raw technical capability than about everything else surrounding it.

Interestingly, change management and board-level support ranked far lower, with scores of 42 and 35 respectively. This suggests that getting buy-in from leadership isn't the bottleneck; the real challenge is making AI work with the messy reality of legacy systems and fragmented data.

Why Are Banks So Eager to Adopt AI Despite These Risks?

Despite these concerns, financial institutions are moving forward with AI adoption at a striking pace. Generative AI, the technology behind tools like ChatGPT, attracted interest from 77% of respondents, making it the most commonly cited AI technology under consideration. Machine learning followed at 68%, while robotic process automation, natural language processing, and agentic AI (AI systems that can take autonomous actions) were cited by 47%, 46%, and 44% of respondents, respectively.

The appeal is clear: AI promises to automate tedious compliance work and catch risks faster than humans can. Among firms considering agentic AI specifically, transaction monitoring emerged as the top priority use case at 66%, followed by fraud detection at 55% and sanctions screening at 46%. Customer onboarding and KYC (know-your-customer) maintenance, which require verifying customer identities and assessing financial risk, were identified by 42% and 41% of respondents.

How to Navigate AI Implementation in Compliance Operations

For financial institutions serious about deploying AI in compliance, the survey results point to several critical priorities:

• Data Governance First: Before implementing any AI system, establish clear data quality standards, audit existing datasets for accuracy and completeness, and create processes to maintain data integrity over time.
• Legacy System Integration Planning: Map out how new AI tools will connect with existing banking infrastructure, identify data silos that need to be bridged, and allocate budget for middleware or data consolidation projects.
• Build Internal Expertise: Hire or train staff who understand both compliance requirements and AI capabilities, create cross-functional teams that include compliance officers and data scientists, and establish clear governance frameworks before deployment.
• Regulatory Engagement: Work with regulators early to understand expectations for AI transparency and explainability, document how AI systems make decisions, and prepare for audits that will scrutinize both the technology and the data feeding it.

The survey identified legacy systems, fragmented data, and manual processes as the primary obstacles making AI implementation more difficult. Beyond these technical challenges, governance, data management, and shortages of staff with required technical skills emerged as critical pain points for compliance teams.

What makes this moment particularly important is the mismatch between ambition and readiness. Financial institutions are committing to AI adoption at scale, but many are doing so without fully addressing the data quality issues that will determine whether these systems actually work. The firms that succeed will likely be those that treat data quality not as a technical problem to solve later, but as a foundational requirement that must be addressed before any AI system goes live.