Building Trustworthy AI: Enterprises Should Focus on AI Readiness and Assurance Systems

Artificial intelligence is helping CFOs and their finance teams improve efficiency and productivity, but the AI opportunities across the broader organization are even greater.

"At a macro level, there are billions of dollars in opportunities waiting to be unlocked," said Bryan McGowan, KPMG U.S. Global and U.S. Trusted AI Lead. "However, how this macro opportunity translates into specific opportunities for each organization will depend on the strategies and methods they use to leverage it."

To unlock maximum value, AI strategies need to be rooted in trust and accuracy.

In finance and other critical business environments, AI tools and processes that merely protect data and produce accurate outputs are not sufficient. Truly useful AI systems must consistently deliver accurate answers, manage data responsibly, and comply with privacy and data protection regulations. Human oversight is essential to ensure reliability and adherence to legal, ethical, risk, and regulatory standards.

Developing a trusted AI strategy, supported by AI readiness and assurance measures, is key to mitigating risks and driving long-term impact with AI.

Starting from a trusted framework

Applying AI in finance and other business contexts introduces risks, from hallucinations, bias, and inaccuracies (specific to individual generative AI solutions), to security risks from data breaches and long-term reputational damage.

As AI adoption surges across existing and new use cases, these risks are expected to grow.

"Overall, finance is still in the early stages of agentic AI workflows and agent-led systems," said Brian Fields, KPMG U.S. Audit Transformation Leader. "A major concern for organizations is that as agents become ubiquitous, managing all these agents, ensuring they operate correctly, and having appropriate security will become very difficult."

Having a trusted AI foundation addresses this concern. In the AI context, trust encompasses considerations such as reliability, security, safety, privacy, sustainability, explainability, data integrity, transparency, fairness, and accountability—which is exactly what KPMG follows internally and in collaboration with clients, as outlined in theTrusted AI framework's ten pillars.

Validating AI claims and controls

One of the key ways organizations can gain confidence in their trusted AI foundation across the ten considerations above is through AI assurance.

"AI assurance involves a third-party provider stepping in to independently verify whether an organization's actions at the governance level and at the application or system level meet the requirements of trusted AI," said Fields.

AI assurance also involves verifying that the controls an organization has put in place around its AI systems operate as intended, and that its claims about governance programs or the security, safety, and reliability of its AI systems are true.

Assurance also provides organizations with the opportunity to obtain third-party certifications, such as theISO/IEC 42001international standard.

Confirming through assurance that financial AI systems have appropriate controls helps CFOs gain confidence that AI-driven efficiency and productivity gains are not coming at the expense of data security or quality, nor exposing the organization to other high risks.

"Most organizations are investing heavily in their AI transformation journey," said McGowan. "Their CFOs are seeking validation: first, that they are transforming with AI in a responsible manner and not exposing the organization to risk; and second, that they are getting value or return on investment from their AI investments, and whether it is worth continuing."

First steps toward trusted AI to drive impact

As organizations advance on their AI journey, whether just starting out or advancing existing AI initiatives, the path to success lies in prioritizing the building of trusted AI. For early adopters, this means starting with controlled experiments, establishing governance structures, and implementing human-in-the-loop controls. For more mature organizations, it requires adopting more sophisticated governance and monitoring systems to maintain effective controls and oversight.

The future of AI in business is clear: a shift toward more autonomous AI systems and AI integrated into workflows. To be part of this future, organizations must act now. They can start by thoroughly testing and validating AI systems, developing governance strategies to guide AI development responsibly, and considering initial investments in AI assurance to enhance confidence in their AI systems.

The potential value of AI is enormous, from increasing productivity and reducing costs to unlocking new revenue streams through data insights. Using AI to extract insights from data and leveraging those insights to identify better ways to engage customers is just one way to unlock new value from data. However, this can only be achieved safely and reliably with trusted AI at its core. As Bryan McGowan stated: "When you look at AI from a value perspective, governance—and by extension, trust and assurance—becomes even more important. It's not just an expense item or a cost lever; it's an opportunity for organizations to improve their AI systems, better understand and measure the value AI brings to the business." By embracing trusted AI, organizations can not only mitigate risks but also drive long-term impact and realize the full potential of their AI investments.

To learn more about KPMG's strategic approach and frameworks for designing, building, deploying, and using AI strategies and solutions in a responsible and ethical manner, please clickhere。