Credible model audit for the AI era

Our new white paper explores why assurance must move beyond good intentions and informal opinions. It sets out the case for a standards-based approach, combining ISAE 3000, the global benchmark for assurance on information other than historical financial statements, with ISQM 1, a firm wide system of quality management. This combination establishes a minimum credible standard for model audits. It clarifies the scope and level of assurance, provides evidence of work performed, embeds quality controls and supports opinions that are both durable and bankable.

Download the white paper

Three forces are making change urgent.

Regulatory accountability is increasingly important.

Regulators expect stronger model governance and oversight. Banks and regulated investors are expected to manage model risk and data security across their supply chains. Assurance aligned to recognised standards helps meet supervisory expectations and protects sensitive information.

Artificial intelligence (AI) is changing the workflow.

Artificial intelligence is now becoming part of both spreadsheet development and review processes. Large language models promise efficiency but can produce plausible yet unreliable outputs on complex spreadsheet logic. Independent, transparent and well documented review is essential to maintain confidence.

Market credibility is at stake.

The title “model auditor” is still unregulated, which creates a credibility gap that lenders, investors and sponsors cannot afford to ignore. There is wide variation in practices and documentation, and occasional provider fragility undermines trust in opinions that are not grounded in recognised standards.

Download the white paper to see how a standards-based approach can close the credibility gap, address the challenges of artificial intelligence and align with evolving supervisory expectations, all without slowing down transactions.

Download the white paper

Extracts from this article previously appeared on Model auditors warn investors of AI risk by the Partnerships Bulletin.