AI's transformative role in financial reporting

This was the central theme of today’s panel discussion at the Conference¹ moderated by Mathew Watson, Senior Vice President & Chief Accounting Officer, Best Buy, which explored how AI strengthens, rather than undermines, the core principles of strong internal controls, investor confidence and transparency.

Below are the key takeaways on the AI-driven evolution of financial reporting.

The deployment of AI within finance processes is not without its challenges. Data security and privacy, reliability and compliance are all critical, and finance leaders must proactively address these issues to mitigate risks. The role of the auditor is thus evolving to include the assessment of AI control environments, so that AI systems operate within established ethical and regulatory frameworks.

Jamie Lynch

KPMG Deputy Chief Auditor

Beyond experimentation: AI in practice today

Successful transformation isn't about deploying the latest technology for its own sake, but about fundamentally solving business problems. The most significant gains come from a human-centric rethinking of how work gets done.

Leading finance organizations are already moving beyond pilots to full-scale deployments. Amie Thuener, VP, Corporate Controller & Principal Accounting Officer at Alphabet, described a dual approach of a bottoms-up ‘AI for all’ culture combined with a top-down mandate to ‘rethink how we work’. Practical applications already in use include:

Automated contract analysis: AI tools that analyze contracts and draft initial accounting memos for complex topics like revenue recognition.
Intelligent expense auditing: AI agents that audit 100% of employee expense reports, a significant leap from the small samples previously performed by humans.
Enhanced analytics: FP&A teams using AI to produce variance analyses complete with written explanations of the key drivers.

This new paradigm demands a human-centric, cross-functional approach that breaks down traditional silos. Only by establishing human owners for end-to-end processes can organizations leverage AI to drive greater accuracy, efficiency and consistency across their operations.

Read our primer on human-centric AI, What’s Your aIQ?

Rethinking controls in an AI-driven environment

While AI introduces new risks like model drift and bias, the foundations of control design remain rooted in COSO concepts. The consensus among panelists is that we need an AI-specific framework, but it will be an evolution, not a wholesale replacement of existing principles.

Thuener outlined five key control approaches that companies can deploy:

Human in the loop: A human expert reviewing and validating AI output.
Performance testing: Using test data with known inputs and expected outputs to validate the model is operating as intended.
Multi-model validation: Employing a second, independent AI model to corroborate the results of the primary model.
Data analytics: Using traditional advanced analytics to monitor the AI system’s activity and outputs to quickly spot anomalies.
Third-party validation: Engaging an independent third party for an assessment of the model's effectiveness.

It is critical that the control approach be tailored to the risk and materiality of the underlying transactions and the objective of the AI model. As confidence in AI models grows, we will likely see a gradual shift from a heavy reliance on ‘human in the loop’ to more automated and AI-centric controls.

To learn more about how solid guardrails can help you scale AI faster, read our article, Essential elements of responsible AI. Our in-depth guide, AI and automation in financial reporting, offers insights for responsibly integrating AI into financial reporting, including governance and internal control considerations.

The conversation around AI in financial reporting must be anchored in governance. Establishing a robust governance framework is not about restricting innovation but enabling it responsibly. At KPMG, AI is integrated into the core of our operations and grounded in our Trusted AI framework, empowering our people to deliver trusted insights and a better experience.

Mamta Soni

KPMG Partner, Department of Professional Practice

The regulator's view: Adapting controls for an AI-driven world

In the earlier conference session on OCA’s current projects, OCA staff emphasized that while AI introduces new complexities, the principles of sound governance and risk management remain foundational.

Nigel James, Senior Associate Chief Accountant, emphasized that successful AI adoption in financial reporting hinges on strong governance and informed oversight. Organizations must set clear policies and ensure boards and committees have the expertise to assess AI-related risks.

Anita Doutt, Senior Associate Chief Accountant, stressed the importance of explainability and audit trails to validate outputs, while recognizing that AI models are dynamic and require continuous monitoring rather than periodic checks. Despite these advances, human judgment remains essential, particularly as new risks like data poisoning and prompt engineering emerge. To navigate this evolution responsibly, OCA continues to monitor AI’s impact and encourages ongoing dialogue across the financial reporting ecosystem.

AI is transforming financial reporting, but the journey is just beginning. Successfully deploying AI in a regulated environment is a journey of collaboration and continuous learning. It demands a culture that encourages responsible experimentation and a workforce equipped with new skills. The future belongs to those who embrace change, experiment responsibly and partner across disciplines to unlock AI’s full potential.

KPMG Private

Relevant Results

Beyond experimentation: AI in practice today

Rethinking controls in an AI-driven environment

The regulator's view: Adapting controls for an AI-driven world

The path forward: Partnership and a new skillset

Subscribe to stay informed

Subscribe to stay informed

Choose your subscription (select all that apply)

Accounting Research Online

Contact KPMG

Headline