KPMG frames Trusted AI as a practical governance framework

KPMG’s Trusted AI framework is meant to decide who can use AI, what gets checked, and how the work is documented when models help with analysis or drafting.

What KPMG is really selling when it says Trusted AI

KPMG has shifted Trusted AI from principle to execution. The firm’s October 2023 PDF, titled "The KPMG Trusted AI approach," carried the subtitle "Accelerating the value of AI with confidence." Its 2024 Australian paper, "Trusted AI governance," framed responsible AI as a way to elevate business and leadership outcomes. The newer global framework page describes the Trusted AI principles as designed to translate responsible AI into practical, scalable action.

KPMG is not just talking about ethics in the abstract. It is trying to create a common language for use cases, approvals and controls so AI can move from experimentation into enterprise deployment without creating privacy, legal or reputational holes.

The day-to-day test: what decisions the framework should answer

For consultants, auditors and advisory professionals, the practical question is not whether Trusted AI sounds good. It is whether it tells you what to do when a manager asks for an AI-assisted draft before a client meeting or when a team wants to use generative AI in an engagement workflow. The framework points to the right categories of control: who owns the data, who approves the use case, how bias is tested, what level of human oversight is required, and how decision-making is documented.

Those are not theoretical questions in a firm like KPMG. If AI is helping draft a memo, summarize a risk assessment or organize audit evidence, staff still need to know whether the output can be used as-is, what must be reviewed, and where the human sign-off sits. In practice, a useful Trusted AI policy would tell employees whether the data can leave the firm’s environment, whether client information can be entered into a third-party tool, and what evidence needs to stay in the file to show the result was validated rather than blindly accepted.

A principles-based framework can say "use responsible AI," but staff still need a plain answer on whether low-risk drafting is treated differently from client-facing analysis, whether model outputs must be cited in workpapers, and whether one partner’s approval is enough or whether legal, risk or privacy review is required for certain use cases.

Why the commercial push matters

KPMG has also started turning Trusted AI into a service line. On May 7, 2025, the firm announced KPMG AI Trust, a suite of services designed to help clients ensure AI reliability, accountability and transparency, enabled by ServiceNow. On September 25, 2025, KPMG LLP expanded those services with new AI Assurance capabilities. Trusted AI is becoming part of the firm’s client-facing advisory and assurance offer.

For people in consulting and advisory, that creates a familiar professional-services dynamic: the firm is selling the discipline it expects to apply to itself. That can be a commercial advantage if KPMG can show it actually uses the same controls in its own work. It can also create skepticism if employees see the messaging outrunning the operating model. The closer the firm gets to AI assurance, the more its own internal rules will be judged against the standards it recommends to clients.

The trust problem behind the framework

KPMG has paired the framework with research on trust. In 2023, KPMG and the University of Queensland published "Trust in Artificial Intelligence: A Global Study," led by Professor Nicole Gillespie, Dr Steve Lockey, Dr Caitlin Curtis and Dr Javad Pool. KPMG’s 2025 global trust-in-AI study found more than half of people globally were unwilling to trust AI.

In October 2025, the firm’s Australian GenAI board survey found that only 3% of Australian organizations had integrated GenAI into strategy. The obstacle is not just access to tools; it is governance. Boards and management teams may see the upside, but many are still waiting for enough control structure to move from pilots to strategy.

What that means inside KPMG

Inside the firm, Trusted AI is becoming a baseline capability rather than a niche specialty. Employees do not need to be machine learning engineers to work under this framework, but they do need to understand data handling, model review, documentation and the limits of automation. That is especially relevant in audit, where the assurance environment depends on evidence, and in consulting, where clients will expect KPMG to explain why a recommendation is both fast and defensible.

KPMG is trying to make AI use reviewable by people who are not technical specialists. A good governance framework gives staff enough structure to ask whether a use case is appropriate, whether the inputs are clean, whether bias or privacy concerns have been checked, and whether the final judgment still belongs to a human.

The open question for employees

The firm has put privacy, controls and regulatory compliance at the center of its AI governance messaging. But the value of a Trusted AI framework will be judged in everyday execution: whether it tells a consultant when to escalate a use case, whether it gives an auditor a defensible standard for AI-assisted analysis, and whether it helps a manager explain to a client why speed did not come at the expense of control.