AI is starting to appear in corporate reporting, but its use remains cautious and uneven.

Recent Financial Reporting Council research found that the use of AI in corporate reporting is increasing, with generative AI being used particularly in narrative reporting. Its use in financial statements remains more limited, and corporate reporting remains strongly human-led.

For finance teams, the practical issue is not whether AI can be used at all. It is whether any AI-supported output, analysis or draft disclosure can be traced back to reliable source data, reviewed by someone accountable and explained clearly if the auditor asks.

AI can support reporting, but it does not change management’s responsibility for the final figures, judgements and disclosures.

You can read the FRC’s research here: Corporate reporting remains human-led amid growing adoption of artificial intelligence.

The FRC also published guidance in March 2026 on the use of generative and agentic AI tools in audit engagements. That guidance is aimed at audit firms. It does not create a new standalone reporting obligation for finance teams, but it does show the wider direction of travel: AI-supported work needs appropriate oversight, documentation and professional judgement.

You can read the FRC’s audit-firm guidance here: Innovative new guidance supports audit firm adoption of emerging AI technologies.

Where AI may appear in finance reporting

AI use in finance teams is not always obvious.

It may sit inside software, spreadsheets, reporting tools, data analytics platforms or general productivity tools. It may be used formally by the finance team, or informally by individuals trying to work more efficiently.

Common examples may include:

• Drafting narrative reporting or commentary.
• Summarising management results.
• Reviewing large data sets.
• Identifying unusual transactions.
• Supporting reconciliations.
• Preparing variance commentary.
• Producing draft board reporting.
• Extracting information from documents.
• Supporting analysis for forecasts or budgets.

These uses may be helpful, but they also create questions. What source data was used? Was the output checked? Was the tool appropriate for the task? Who reviewed the result? Can the business explain how the final figure or disclosure was reached?

These questions matter before the audit begins, not after the auditor raises them.

AI support is not the same as reliable evidence

An output that looks plausible is not the same as one that is correct.

That is one of the main risks when AI is used in reporting. A draft paragraph may read well but include a statement that is not supported. A summary may miss an important exception. An analysis may rely on incomplete or unreconciled data. A tool may produce an answer without showing enough of the process behind it.

Finance teams should therefore treat AI-supported outputs as starting points, not final conclusions.

Before relying on them, management should be able to show:

• What tool was used.
• What information was input.
• Whether the data was complete and reconciled.
• Who reviewed the output.
• What checks were performed.
• What changes were made before the final version was approved.
• Who owns the final figure, analysis or disclosure.

The aim is not to prevent useful technology from being used. It is to make sure the business can trust and explain what is being reported.

What auditors may ask

Auditors are unlikely to object to technology simply because it has been used. The issue is whether the output can be evidenced and whether management has applied appropriate review.

If AI has supported any part of the reporting process, auditors may ask questions such as:

• Where has AI been used in preparing the accounts, reports or disclosures?
• Was the use of AI approved or informal?
• What source data was used?
• Was that source data reconciled to the accounting records?
• Who reviewed the AI-supported output?
• Was the output checked against underlying evidence?
• Were any errors, omissions or inconsistencies identified?
• Has management retained an audit trail?
• Who approved the final version?

These questions are not only technical. They go to the reliability of the reporting process and the strength of management oversight.

Data quality, transparency and explainability

AI is only as useful as the information and controls around it.

If the source data is poor, incomplete or unreconciled, the output may be unreliable. If the tool’s process cannot be understood or explained, it may be difficult to evidence. If no one owns the final review, accountability becomes unclear.

The FRC’s work on AI in corporate reporting highlights risks around error, data quality, transparency and explainability. These are practical issues for finance teams, especially where reporting needs to withstand audit, lender review, investor scrutiny or buyer due diligence.

A good process should therefore make clear:

• Which data sources are being used.
• How those data sources are checked.
• Whether the output can be explained.
• Whether limitations of the tool are understood.
• Whether sensitive or confidential information is being handled properly.
• Whether the final judgement remains with management.

Technology may speed up reporting, but speed is only valuable if the output remains reliable.

Controls finance teams should document

A sensible control process does not need to be overly complicated.

It should give management and the auditor a clear view of where AI is used, what it produces and how the output is reviewed.

Useful controls may include:

• A simple register of where AI tools are used in reporting.
• Clear rules on what AI can and cannot be used for.
• Review and approval steps for AI-supported outputs.
• Reconciliation of source data before it is used.
• Evidence of checks performed on draft disclosures or analysis.
• Version control for narrative reporting.
• Clear ownership of final numbers and statements.
• Escalation routes where the output appears inconsistent or unsupported.
• Periodic review of whether tools remain appropriate.

The aim is confidence in the reporting process, not paperwork for its own sake.

Why human review remains central

AI can assist with analysis, drafting and summarising information, but it does not replace professional judgement.

Finance reporting often involves judgement about provisions, impairment, going concern, revenue recognition, estimates, assumptions, disclosures and the presentation of performance. These areas require management to understand the business, assess evidence and take responsibility for the conclusion reached.

Human review is especially important where:

• The area involves judgement or estimation.
• The output is used in statutory accounts.
• The wording affects disclosures to shareholders, lenders or regulators.
• The data comes from multiple systems.
• The business is preparing for funding, investment or sale.
• The board needs confidence in the reporting process.

A finance team should not rely on an AI-supported output simply because it appears fluent, detailed or consistent. The question is whether it is right, complete and supported by evidence.

Questions to ask before the audit

A short internal review before audit fieldwork can reduce difficulty later.

Useful questions include:

• Do we know where AI is used in our reporting process?
• Is that use approved, documented and understood?
• Are AI-supported outputs reviewed by someone accountable?
• Is the source data reliable and reconciled?
• Can we explain how key outputs were produced?
• Have draft disclosures been checked for accuracy and balance?
• Are any judgemental areas still clearly owned by management?
• Is there an audit trail showing what was reviewed and approved?
• Would the board be comfortable explaining the process to the auditor?

These questions help move AI use from an informal productivity tool to a controlled part of the reporting process.

How this supports better decision-making

It would be easy to treat this as an audit-readiness task.

The more useful view is that strong oversight of AI in reporting improves the reliability of the information management itself depends on.

Working through the process can highlight weaknesses in reporting, gaps in reconciliations, unclear ownership and inconsistent controls. Reporting that is well controlled, whatever tools support it, gives the board firmer ground for its decisions and greater credibility with lenders, investors and any future buyer.

AI can support the reporting process, but it does not remove management’s responsibility for the figures, narrative and judgements. Using AI does not, on its own, make reporting more reliable.

Strong controls should not create unnecessary bureaucracy. The aim is output that can be trusted and explained.

Final thought

AI in reporting is, at heart, a question of oversight: knowing where it is used, checking what it produces and owning the result.

For most finance teams, the sensible response is to map where AI tools sit in the reporting process, make sure outputs are reviewed and evidenced, and keep clear ownership of the final numbers and disclosures.

Handled well, AI can support reliable reporting. Handled informally, it can create a question mark over figures and narrative that management still remains responsible for.

Accendo supports growing businesses and organisations with audit readiness, financial reporting and advisory work where reliable evidence, clear documentation and practical judgement matter. Where AI is used in reporting, finance teams should be able to show what was used, how outputs were reviewed and who remains responsible for the final figures and disclosures. 

This article is for general information only and reflects the FRC position at the date of publication. It should not be treated as legal, accounting, tax, transaction, investment or governance advice. You should obtain specific advice based on your organisation’s circumstances.