Artificial intelligence can produce a polished paragraph in seconds. It can summarise performance, identify trends and explain movements in revenue, costs or cash flow.
That does not mean it understands the business.
An AI tool may be able to draft sections of an annual report, but the board, finance team and individual accountants remain responsible for the judgements behind the words. They must decide whether the information is accurate, balanced and consistent with the financial statements. They must also challenge explanations that sound convincing but are not supported by evidence.
This creates an important issue for anyone studying SBR ACCA. AI is not simply a technology topic. It is a reporting, governance and ethics topic that connects professional judgement, accountability, internal control and the quality of corporate communication.
Candidates who can explain those connections clearly will be better prepared for current issues requirements and broader questions involving boards, audit committees and professional accountants.
For wider support on applying reporting principles to exam scenarios, an experienced ACCA SBR tutor can help candidates turn technical knowledge into focused, professional answers.
The annual report is more than a writing exercise
It is easy to think of an annual report as a large document that needs to be assembled, checked and published.
That view is too narrow.
The annual report is one of the main ways a company explains its performance, financial position, risks, strategy and governance to investors and other stakeholders. Its purpose is not simply to contain the correct words. It must communicate a fair and understandable picture of the business.
AI may help produce those words, but it cannot own the message.
Management remains responsible for the information included in the report. Directors must be satisfied that the report is consistent with their understanding of the company. The finance team must ensure that narrative claims agree with the accounting records, forecasts and financial statements.
That responsibility does not disappear because software produced the first draft.
In fact, the use of AI may increase the need for review because the output can appear confident even when the underlying conclusion is weak.
The difference between drafting and deciding
AI can draft a paragraph explaining that revenue increased because of strong customer demand.
The difficult question is whether customer demand was actually the cause.
Perhaps revenue increased because prices were raised. Perhaps the company acquired another business. Perhaps sales increased but cash collection deteriorated. Perhaps demand was strong in one market but falling sharply in another.
The words may sound reasonable while hiding the real drivers.
Professional judgement begins where automated drafting ends.
The accountant must identify the evidence, understand the commercial context and decide which explanation is accurate. The final report should reflect the substance of the company’s performance rather than the easiest narrative to generate.
This distinction is useful in an SBR answer.
A weak answer might say that AI makes reporting faster.
A stronger answer would explain that AI may improve efficiency, but management must retain responsibility for selecting, verifying and presenting information. It should also discuss the risks created when a plausible draft is accepted without sufficient challenge.
Who is responsible when AI gets it wrong
Responsibility should not become unclear simply because several people and systems contributed to the report.
The board remains responsible for the annual report. The chief financial officer and finance team remain responsible for the financial information they prepare. The audit committee remains responsible for overseeing the reporting process and challenging significant judgements.
The external auditor has a separate responsibility to obtain sufficient appropriate evidence and form an independent opinion. The auditor cannot simply rely on the fact that management used an advanced system.
An AI supplier is not the person signing the financial statements.
This is a central governance point.
Companies need clearly assigned ownership at every stage of the reporting process. Someone must approve the data used by the tool. Someone must review the generated output. Someone must confirm that changes have been made correctly. Someone must provide final approval.
Without clear ownership, errors can pass through the process because each person assumes another person has checked them.
Why polished output can be dangerous
Poor writing is usually easy to spot. Polished AI output can be more dangerous because it creates an impression of authority.
A paragraph may be grammatically perfect, logically structured and completely wrong.
The danger is greatest where the underlying issue requires judgement. This could include impairment assumptions, going concern assessments, provisions, climate-related risks, useful lives, revenue recognition or the classification of an arrangement.
AI may summarise the relevant accounting rules, but the correct treatment depends on the specific facts.
For example, software may explain the difference between a joint operation and a joint venture under IFRS 11. It cannot make a reliable classification unless it has complete and accurate information about the legal structure, contractual terms and the parties’ rights and obligations.
A confident answer created from incomplete information is still an unreliable answer.
The annual report must agree with the numbers
One of the biggest risks is inconsistency between the narrative report and the financial statements.
The strategic report might describe demand as strong while the impairment model assumes a significant fall in sales. The risk section might describe climate exposure as limited while management has reduced asset lives or increased provisions because of environmental pressures.
A company might celebrate improved profitability while operating cash flow has weakened significantly.
AI can make each section read well independently. The finance team must ensure that the report makes sense as a complete document.
This is sometimes described as connectivity.
The narrative, assumptions, accounting treatments, alternative performance measures and financial statement disclosures should tell a consistent story. Differences may be legitimate, but they need to be understood and explained.
In an SBR answer, candidates should not discuss AI only as a drafting tool. They should connect it to the risk of inconsistency across the annual report.
The role of professional scepticism
Professional scepticism is often discussed in relation to auditors, but the underlying attitude is also valuable for preparers.
A finance professional should not accept an AI-generated explanation simply because it sounds reasonable. The output should be challenged.
What data was used?
Is the information complete?
Does the explanation agree with the accounting records?
Has contradictory evidence been ignored?
Are the conclusions consistent with board papers and forecasts?
Has the wording become more positive than the underlying evidence justifies?
These questions turn passive review into active challenge.
Professional scepticism does not mean assuming every AI output is wrong. It means avoiding uncritical acceptance.
AI does not remove management bias
Management bias existed long before artificial intelligence.
Companies naturally want to present themselves positively. Difficult results may be described as temporary. Strong results may be attributed to strategy, while weak results are blamed on external conditions.
AI can amplify this problem if it is prompted to make a report sound more confident, reassuring or investor-friendly.
The tool may soften negative language, remove uncertainty or emphasise favourable information. It may create a narrative that is technically accurate at sentence level but unbalanced when considered as a whole.
The board and audit committee must therefore consider whether the report gives appropriate prominence to negative as well as positive information.
A balanced annual report does not hide problems. It explains them clearly and gives users enough information to understand management’s response.
What good AI governance should include
Companies do not need to ban AI from the reporting process. They need controls that reflect the risk of the task.
A useful governance framework should cover:
- approved tools and permitted uses
- protection of confidential information
- control over the data entered into the system
- clear ownership of generated content
- evidence supporting important statements
- independent review of significant judgements
- version control and records of material changes
- final approval by appropriately senior people
These controls should be proportionate.
Using AI to improve the wording of a routine internal paragraph creates a different level of risk from using it to draft a going concern assessment or describe a major accounting judgement.
The more significant the subject, the stronger the human review should be.
Confidentiality cannot be an afterthought
Annual report preparation involves sensitive information.
Finance teams may work with unpublished results, forecasts, acquisition plans, legal disputes, restructuring proposals and personal data. Entering this information into an unsuitable AI platform could create a serious confidentiality risk.
The fact that a tool is convenient does not mean it is approved.
Companies need policies that explain which systems may be used, what information may be entered and how outputs should be stored. Staff also need training because controls are ineffective when people do not understand the reason for them.
This is a strong ethics point for an SBR answer.
The professional accountant has a duty to protect confidential information. That responsibility continues when new technology is introduced. Convenience does not override professional obligations.
Competence means understanding the limitations
Professional competence in the AI era does not mean every accountant must become a software developer.
It does mean accountants need enough understanding to use the technology responsibly.
They should know that AI output can contain errors, omit important information or create unsupported conclusions. They should understand that the quality of the output depends heavily on the data and instructions provided.
They must also recognise when specialist support is needed.
An accountant who uses AI without understanding these limitations may place excessive reliance on the tool. That can weaken professional judgement rather than support it.
The correct approach is to use technology as an assistant, not as a substitute for competence.
The audit committee should ask better questions
The audit committee does not need to review every prompt entered into an AI system.
It should, however, understand how AI is being used in the reporting process and whether the related risks are controlled.
Useful questions include whether AI has been used in preparing significant narrative sections, how generated content is checked, who approves the output and how confidential data is protected.
The committee should also ask whether management has tested the consistency of the narrative with the financial statements and other information available to the board.
This oversight is particularly important when the technology is introduced quickly. A reporting process may change before the formal control environment catches up.
The external auditor cannot outsource scepticism
AI may also be used by auditors to analyse data, identify unusual transactions or assist with documentation.
The same principle applies. The auditor remains responsible for the audit opinion and the judgements supporting it.
An AI-generated analysis does not automatically become audit evidence. The auditor must understand the source and reliability of the information, evaluate the tool’s output and investigate matters that require further work.
There is also a risk of automation bias. This occurs when people place too much confidence in a system and become less likely to challenge its conclusions.
A good SBR or ethics answer should therefore avoid claiming that AI automatically improves audit quality. It may improve the speed or scope of certain procedures, but the benefit depends on governance, testing, review and professional scepticism.
How this topic could appear in an SBR exam
AI may appear as a direct current issues topic, but it could also be built into a wider scenario.
A company might use generative AI to prepare narrative disclosures. An audit committee may be concerned about inaccurate information. A finance employee might enter confidential data into a public system. An auditor might rely on an automated analysis without understanding how it was produced.
The requirement could ask candidates to discuss ethical issues, governance weaknesses, reporting risks or the actions the board should take.
The best answers would not spend several paragraphs defining artificial intelligence.
They would identify the issue, connect it to the scenario and explain a practical response.
A strong structure for an exam paragraph
A useful paragraph structure remains:
Issue – Principle – Application – Recommendation.
For example:
Issue – Management has used an AI tool to draft the impairment disclosure without documenting how the output was checked.
Principle – Directors remain responsible for the annual report, while professional accountants must exercise competence, care and appropriate judgement.
Application – The generated wording may not reflect the assumptions used in the impairment model and could create inconsistency between the narrative disclosure and the financial statements.
Recommendation – A suitably experienced member of the finance team should verify the disclosure against the approved model, document the review and submit significant judgements to the audit committee.
That paragraph is short, applied and professional.
It also demonstrates why candidates need more than technical knowledge. They need to communicate the consequences and recommend a realistic action.
Avoid generic answers about technology
A weak answer often says that AI is fast but may make mistakes.
That is too general.
A stronger answer identifies the particular reporting risk.
The problem might be an unsupported claim, incomplete data, confidentiality, management bias, inconsistent disclosure, unclear responsibility or insufficient review.
The safeguard should then respond directly to that risk.
For example, training is not the answer to every problem. If the risk is that no one owns the final disclosure, the safeguard is clear approval responsibility. If the risk is confidential data being entered into a public tool, the safeguard is an approved system and restricted information policy.
Specific problems need specific responses.
Judgement becomes more valuable rather than less
Some candidates worry that AI will reduce the value of accounting knowledge.
The more likely change is that judgement becomes more valuable.
When information can be generated quickly, the important skill is deciding whether it is useful, reliable and relevant. Accountants must understand the reporting framework, the business and the interests of users well enough to challenge the output.
They must also explain uncertainty honestly.
AI can help organise the work, but it cannot take professional responsibility. It cannot appear before an audit committee and defend a judgement. It cannot sign the financial statements or accept the consequences of misleading reporting.
That is why the human role remains central.
How candidates should prepare for this type of question
Candidates should practise discussing AI through established SBR themes rather than treating it as a separate technical syllabus.
Connect it to:
Governance and board responsibility.
Ethical principles and confidentiality.
Professional competence and scepticism.
Internal controls over reporting.
Consistency between narrative and financial information.
Audit evidence and documentation.
Management bias and balanced communication.
This approach makes the topic easier because the underlying principles are familiar.
Candidates who want structured practice applying current reporting issues to scenario-based questions may benefit from an ACCA SBR course that includes timed writing and feedback.
The board must own the final message
AI can help a company produce an annual report more efficiently. It may help teams organise information, identify inconsistencies and improve readability.
Those benefits do not change the ownership of the report.
The board must own the judgements. The finance team must own the accuracy of the financial information. The audit committee must challenge the process. The auditor must maintain independent professional scepticism.
A company should therefore be able to explain not only where AI was used, but also how the output was controlled and who approved it.
The most important question is not whether AI helped write the annual report.
It is whether the people responsible for the report understood, challenged and stood behind every significant conclusion.
That is where professional judgement begins.
