Client
Global Asset and Wealth Management Firm
Industry
Capital Markets
Primary goal
AI system card implementation
Technologies
Generative AI
Make AI systems you can trust
Under the leadership of its Chief Data Officer, the client sought to strengthen how it assessed and governed AI systems by introducing a more robust and repeatable testing approach across its technology environment. The client needed support in addressing gaps in its technology review process and building a more scalable model for testing internally developed and third-party AI systems before production deployment. The goal was to improve trust, accountability and regulatory compliance aligned with the company’s internal AI governance policies and external regulations.
Client transformation journey
- Before
- During
- After
Lack of transparency
The organization had deployed multiple AI solutions and sought to strengthen how it assessed and governed AI systems. The goal was to develop a more structured method for evaluating AI systems and documenting risks, controls, and readiness criteria by providing a transparent way to demonstrate compliance with stakeholders’ expectations, proactively address potential risks and build confidence in their use.
Design and delivery of an AI system card and governance model
KPMG in the US used a phased approach to design and implement an AI testing framework, starting with measurable metrics to evaluate the performance, safety and transparency of the client’s existing AI system along with test cases to evaluate the company’s compliance with these attributes. A quantitative rubric was developed to evaluate results of tests, and various test steps were automated with technology.
An AI system card template was developed to document how the AI system was built, what it can do, its limitations, data considerations, performance against responsible frameworks and potential risks that the company and its users should be aware of.
Increased transparency and enhanced risk management
The client instituted a requirement, originating from the CEO, that all high-risk AI systems must undergo KPMG’s AI testing before production deployment. AI system cards have improved transparency for the client’s technical and non-technical stakeholders, while also driving faster and more effective risk assessments and evaluation of guardrail effectiveness.
Stronger alignment with internal governance policies and external regulations has also increased the potential for the organization’s AI governance practices to serve as a market differentiator and competitive advantage by moving from point-in-time AI reviews toward a more embedded and scalable governance model integrated into the development lifecycle.
How we make the difference
Across industries, organizations are racing to deploy generative and agentic AI, often without clear governance, defined accountability or alignment to existing control frameworks. KPMG in the US team’s introduction of AI system cards enabled faster, more effective AI testing, risk assessments and increased trust and accountability in the client’s deployment of AI systems.
Please reach out to us if you’re ready to move to a structured, scalable AI implementation designed to support trusted AI use. We welcome the opportunity to show you how system cards can create stronger control environments with transparent risk assessments that give you confidence in scaling AI in a trusted way.
KPMG. Make the Difference.
Meet the team
Some of the imagery/videos were developed with support of AI technology.
Make the Difference with KPMG
At KPMG, what sets us apart isn't just what we do, it's how we do it.
We’re leveraging deep expertise and innovative technology to turn insights into opportunity and move business and innovation forward, placing human connections at the center.
Together, we make the difference.