CW survey: Compliance teams struggling with AI implementation and trust issues

Of the 193 compliance, ethics, risk, and audit leaders surveyed, problems with data quality or data access were cited as a challenge in AI tool implementation by 66 percent of respondents. Next was lack of expertise, at almost 54 percent, and then poor integration with current systems (49 percent). The results suggest that compliance professionals’ challenges with AI tools span the entire process, from initial implementation through the quality and reliability of AI outputs. 

Ensuring data quality was the top challenge for respondents, a result which likely plays a part in an underlying lack of trust that some compliance professionals have with AI tools. Less than half of those responding (42 percent) trusted the outputs they saw from AI tools, while 48 percent were neutral to them, according to the survey. 

Mohan Krishna, konaAI’s Executive Director, Head of Product Innovation, said this result reflects a need for explainability and assurance.

”Compliance teams see AI as inevitable and necessary, but they are constrained by governance risk, data quality, regulatory uncertainty, and skills gap,” Krishna said. ”This is creating a strong demand for enterprise-grade, agentic AI solutions that confidently assist in transaction monitoring, flagging meaningful risks for human review, evaluating contracts, and measurably improving the business, not just generic, publicly available tools like OpenAI to answer questions.”

The primary risks of AI in compliance are not about whether organizations should use these tools, the survey found, but whether they are prepared to use them well. Implementation of AI tools, from both the employee and technology perspectives within a company, was also a challenge for compliance professionals. Training needs for AI tools (47 percent) and unknown or unmanaged use by employees (42 percent) were other pain points they cited. 

Nearly 16 percent of respondents also flagged that employee resistance to the AI tools was an issue. Couple these challenges with the poor integration of current technology, and it becomes clear the difficulties many organizations face in using AI on platforms never designed to support it–especially for employees with minimal training on the technology. As adoption continues to grow, the survey suggested that investments in data governance, integration, training, and clear policies may matter more than the choice of AI technology itself.

Another major theme among the most common challenges for AI tool implementation centered on compliance and regulatory concerns. These included a lack of or inconsistent AI-related policies (29 percent), ethical or bias concerns (27 percent), regulatory uncertainty (27 percent), and a lack of transparency or explainability (25 percent).

Compliance professionals are struggling to understand the policy guardrails for AI use that are central to their function, particularly when those policies are limited, unclear, or, in the case of federal U.S. regulators, non-existent. (A few states, like California and Colorado, have approved AI use regulations; many other states are considering legislation).

With so many issues involving AI-tools, it’s clear compliance professionals are playing catch-up on new technology that is being pushed on them. The source of that push, according to the survey, comes from the executive level. 

“Executive leadership (48 percent) and boards (15 percent) are the primary adoption drivers of AI initiatives according to the survey, not necessarily the grassroots compliance team,” Krishna said. ”This indeed indicates top-down pressure to modernize compliance or risk getting left behind in the company.” 

The survey findings show a growing gap between the speed at which AI is being adopted and the readiness of compliance professionals who are tasked with overseeing its implementation. AI’s expansion in compliance may continue at a rapid pace, but so will the introduction of risk.