By 
Artificial intelligence can make corporate reporting cleaner, faster and more consistent.
It can improve a clumsy paragraph. It can summarise a year of performance. It can compare disclosures with the previous annual report and identify language that has changed. It can even produce a convincing explanation of why revenue increased or margins fell.
What it cannot do automatically is make that explanation authentic.
That distinction is becoming more important as AI moves into corporate reporting.
Recent UK research into the use of AI in reporting found growing adoption of generative AI, particularly for lower-risk drafting and editing tasks. At the same time, companies remain cautious about using it for areas involving significant judgement. Accuracy, accountability and trust remain important concerns, but another issue is emerging alongside them.
Investors still value the sense that the annual report genuinely reflects what management thinks.
That creates an interesting problem.
If every company gains access to increasingly capable writing technology, polished corporate language becomes cheap. A well-written paragraph is no longer evidence that management has thought deeply about the issue behind it.
The valuable part becomes the judgement, specificity and honesty underneath the words.
For ACCA SBR candidates, this is a useful current reporting issue because it connects AI with governance, professional judgement, accountability, investor communication and ethics. Candidates developing those skills with an ACCA SBR tutor should think beyond whether AI can produce accurate wording.
The harder question is whether the reporting still sounds like it belongs to the company.
Corporate reporting has never been only about correct sentences
An annual report has to be technically correct.
Figures should agree with the accounting records. Disclosures should comply with the applicable reporting requirements. Material information should not be omitted.
But technical correctness is only part of useful reporting.
Investors also want to understand what happened to the business.
Why did margins fall?
Why did management continue investing in a weak division?
Why has the company changed its strategy?
What went wrong with an acquisition?
Which risks worry the board most?
What assumptions support the outlook?
These are not simply compliance questions.
They require management to explain its view of the business.
An annual report can contain perfectly accurate sentences while giving very little insight into what management actually thinks.
AI could make that problem worse if companies use it primarily to make reporting smoother rather than more informative.
The danger is not bad writing
The most obvious concern about AI-generated content is that it may contain an error.
That risk is real.
But the more subtle problem is content that is technically acceptable, beautifully written and almost completely uninformative.
Consider this type of statement:
“The group continued to operate successfully in a challenging macroeconomic environment while maintaining its focus on long-term sustainable growth.”
There may be nothing factually wrong with that sentence.
There is also very little worth knowing.
What was challenging?
What happened to demand?
Which costs increased?
Did margins fall?
Did customers leave?
Was growth achieved?
What did management change?
Generic language allows management to describe difficult circumstances without explaining them.
AI is particularly good at producing language of this type because it can create fluent corporate prose from limited information.
The danger is that reporting becomes more professional in appearance while becoming less specific in substance.
Authenticity comes from specificity
Authentic reporting does not require informal language.
It does not mean directors need to write the annual report personally or fill it with opinions.
It means the report should contain information that could realistically belong only to that organisation.
A useful test is simple.
Could the paragraph be copied into a competitor’s annual report without changing anything?
If the answer is yes, it probably lacks specificity.
Imagine a retailer whose operating profit has fallen because distribution costs rose and two new stores performed substantially below expectations.
A generic explanation might say that profitability was affected by inflationary pressures and investment in the group’s growth strategy.
That may be true.
A more authentic explanation would identify the distribution cost increase, explain that the new stores have not reached expected sales levels and describe what management is doing in response.
The second version gives investors something they can evaluate.
Specificity creates accountability.
Once management states what went wrong and what it intends to do, investors can return next year and assess whether the response worked.
AI can imitate tone but not accountability
A generative AI system can learn the style of previous annual reports.
It can produce language that sounds cautious, confident, formal or investor-friendly.
It can mimic the company’s existing tone.
What it cannot do is accept responsibility for the claim.
If management states that a restructuring programme will improve margins next year, management owns that expectation.
If the board says an acquisition remains strategically important despite underperformance, the board should be able to explain why.
If directors describe climate risk as manageable, that statement should be consistent with the assumptions used elsewhere in the financial statements.
AI can draft the words.
The people approving those words still own the judgement.
That is where authenticity and accountability become connected.
Authentic reporting is not simply reporting that sounds human. It is reporting that reflects conclusions people within the organisation are prepared to defend.
Investors can tell when difficult issues are being avoided
One of the strongest tests of authentic reporting is how a company discusses bad news.
Positive reporting is easy.
Management can talk about growth, investment, resilience, innovation and opportunity.
The real test comes when something has gone wrong.
An acquisition underperformed.
A major customer left.
An impairment loss was recognised.
A product launch failed.
A regulatory investigation began.
Cash generation weakened.
An important project was delayed.
These situations create pressure to soften the language.
AI can make that easier.
A prompt asking for a “more positive” or “more reassuring” version of a paragraph may transform a direct explanation into polished language that reduces the apparent seriousness of the issue.
The individual sentences might still be technically correct.
The overall impression may become misleading.
That is why human review cannot focus only on factual accuracy.
Someone needs to ask whether the emphasis is fair.
Authenticity requires bad news to receive appropriate prominence
Balanced reporting does not mean giving exactly the same amount of space to positive and negative information.
It means the importance of the information should determine its prominence.
If a major acquisition has failed to deliver expected returns, the annual report should not bury that fact beneath several pages describing long-term strategic potential.
If operating cash flow has deteriorated materially, a discussion dominated by revenue growth may give users an incomplete picture.
If management has reduced forecasts because of climate-related disruption, the sustainability section should not describe climate exposure as insignificant without explaining the apparent contradiction.
The board needs to ask whether the annual report tells the same story that management sees internally.
That is a powerful authenticity test.
Would an investor reading the annual report reach a broadly similar understanding of the company’s position to someone reading the important board papers?
The two documents will obviously contain different levels of detail.
The underlying story should not be fundamentally different.
The annual report should not become a collection of independently generated sections
Large annual reports are already assembled by multiple teams.
Finance prepares financial information.
Sustainability teams provide environmental data.
Human resources contributes workforce information.
Risk teams prepare principal risk disclosures.
Investor relations may shape performance commentary.
Company secretarial teams contribute governance sections.
AI may increasingly assist each of those teams.
This creates a new connectivity problem.
Every individual section could be well written while the complete report becomes inconsistent.
The sustainability section might describe significant transition risks while the impairment forecasts assume no related effect on future cash flows.
The strategic report might celebrate an acquisition while the goodwill impairment disclosure shows rapidly declining headroom.
The chief executive’s review may describe strong customer demand while receivables have increased sharply and operating cash flow has deteriorated.
AI can improve each section independently.
It takes human judgement to recognise that the sections do not tell the same story.
Authenticity depends on connectivity
This is particularly relevant to SBR.
Candidates should already be looking for contradictions between narrative information and financial reporting assumptions.
An exam scenario might state that management describes a market as highly promising while simultaneously reducing forecasts for the assets operating within that market.
That should trigger challenge.
It may not mean either statement is wrong.
Management may have a reasonable explanation.
Perhaps short-term weakness is expected before a longer-term recovery.
The reporting should explain that distinction.
A strong SBR answer identifies the inconsistency and recommends that management ensures the narrative reporting and accounting assumptions are aligned or clearly explains why they differ.
That is more useful than simply saying the annual report should be accurate.
AI creates a risk of corporate sameness
There is another issue that receives less attention.
As more companies use similar AI systems to draft corporate reporting, language may begin to converge.
Every business becomes resilient.
Every market becomes challenging.
Every strategy becomes disciplined.
Every transformation becomes significant.
Every investment becomes targeted.
Every future becomes one of sustainable long-term growth.
The words sound professional, but their informational value declines.
This is not entirely an AI problem. Corporate reporting contained boilerplate language long before generative AI existed.
AI can, however, make boilerplate easier to produce at scale.
The solution is not deliberately making reports badly written.
It is forcing the drafting process back towards evidence.
Instead of asking an AI system to “write a paragraph about our strong performance”, the reporting team needs to identify the evidence first.
What actually happened?
Which measures changed?
Why?
What evidence supports management’s explanation?
What uncertainty remains?
Once those questions have been answered, AI may help communicate the information more clearly.
The judgement must come before the drafting.
Management voice should come from management thinking
Some companies may worry that using AI automatically makes an annual report less authentic.
That is too simplistic.
Human-written reporting can be generic.
AI-assisted reporting can be specific, balanced and useful.
The important issue is where the thinking happens.
If management has already reached a clear conclusion and AI is used to help express it concisely, the underlying judgement remains human.
If AI is asked to decide what the company should say about a complicated performance issue and management simply approves the result, authenticity is much weaker.
This creates a sensible principle for reporting teams.
Use AI to help communicate decisions.
Do not use it to avoid making them.
Human review needs to ask better questions
Traditional proofreading asks whether the wording is correct.
AI-assisted corporate reporting needs a deeper review.
The reviewer should consider:
- Does this paragraph contain information specific to the company?
- What evidence supports the explanation?
- Does it agree with the financial statements and internal forecasts?
- Has important negative information been softened or omitted?
- Would management defend this wording if challenged directly by an investor?
- Does the section reflect the view expressed internally to the board?
- Has AI improved clarity, or merely made an uncertain conclusion sound more confident?
Those questions address authenticity rather than grammar.
They also make human review meaningful.
A manager who simply reads the generated wording and approves it is technically “in the loop”, but that does not necessarily provide effective oversight.
The reviewer needs to challenge the substance.
Authenticity matters when explaining estimates
Accounting estimates are particularly vulnerable to polished but generic language.
Impairment disclosures are a good example.
A company might state that forecasts reflect management’s best estimate of future market conditions and that appropriate sensitivity analysis has been performed.
That sounds reasonable.
Investors need more.
Which assumption creates the greatest uncertainty?
What has changed since last year?
How close is the asset to impairment?
Has the business historically achieved similar forecasts?
What contradictory evidence has management considered?
These details make the disclosure useful.
AI may help organise the information, but management needs to identify which judgements genuinely matter.
The same principle applies to provisions, going concern, expected credit losses, useful lives and fair value measurements.
Authenticity becomes especially valuable where the accounting depends heavily on management’s view of the future.
There is an ethical dimension too
Professional accountants should not use AI to make reporting technically true but deliberately misleading in tone.
Integrity requires more than avoiding an outright false statement.
Imagine management asks the finance team to use AI to make a weak trading update sound more positive.
The resulting text removes references to customer losses, describes falling demand as temporary and emphasises several favourable indicators.
Every remaining sentence might be factually correct.
The overall communication could still create an unjustifiably positive impression.
The accountant should challenge that approach.
A professional response would explain that reporting should be balanced and supported by evidence. Material negative information should not be obscured simply because software makes it easy to rewrite the narrative.
That is an excellent SBR ethics point because it connects technology with integrity, objectivity and professional behaviour.
The audit committee has a role in protecting authenticity
Audit committees traditionally focus heavily on significant accounting judgements, estimates, controls and audit matters.
As AI becomes more widely used in reporting, they may also need to understand how narrative information is being produced.
That does not mean reviewing every AI prompt.
It means knowing where AI is used, which sections involve significant judgement and who owns the final content.
The committee should be particularly interested in high-judgement areas.
Descriptions of financial performance.
Going concern commentary.
Major acquisition performance.
Climate-related risks.
Significant accounting judgements.
Alternative performance measures.
Principal risks.
These are areas where polished language can influence investor interpretation.
The committee’s role is not to make the report sound less professional.
It is to ensure professional presentation has not replaced honest explanation.
What this means in an SBR exam
Candidates should avoid writing that “investors prefer human-written annual reports”.
That would oversimplify the issue.
The stronger point is that users value accountability, trust and reporting that genuinely reflects management’s view.
AI can assist with drafting while those qualities remain human-led.
A scenario might describe a company using generative AI to rewrite performance commentary.
Management may have instructed the tool to make the wording more positive.
The generated report might then emphasise revenue growth while ignoring weaker cash generation or deteriorating margins.
A strong answer would identify the risk that the narrative is unbalanced.
The candidate could explain that directors remain responsible for the annual report and should verify whether the commentary fairly reflects the underlying performance.
Management should revise the disclosure to explain both the positive and negative developments and ensure the narrative is consistent with the financial statements.
That answer is specific, applied and professional.
Avoid turning authenticity into a vague concept
The word authenticity can sound subjective.
In reporting, it can be made practical.
Authentic reporting should be company-specific.
It should be evidence-based.
It should acknowledge material uncertainty.
It should connect with the numbers.
It should reflect genuine management judgements.
It should provide appropriate prominence to difficult information.
It should leave clear human accountability for the final message.
Those characteristics can be assessed.
This prevents authenticity becoming another vague reporting buzzword.
The better AI becomes the more judgement matters
There is an interesting paradox here.
As AI becomes better at producing professional corporate language, the language itself becomes less valuable.
The scarce resource becomes credible judgement.
Almost any company may soon be able to generate a polished explanation of its results within seconds.
Not every company will be equally willing to explain why an acquisition failed.
Not every management team will clearly acknowledge that a strategy needs to change.
Not every board will give investors useful information about uncertainty rather than hiding behind standard wording.
Those choices remain human.
That may be where the greatest value sits.
What SBR candidates should practise
Candidates do not need to become experts in artificial intelligence.
They need to become better at challenging reporting.
Take a scenario and ask what management is really claiming.
Look for evidence that supports the claim.
Look for information that contradicts it.
Check whether the financial statements tell the same story.
Then write a recommendation that improves the reporting.
This type of exercise develops judgement far more effectively than memorising a generic paragraph about the advantages and disadvantages of AI.
Candidates following an ACCA SBR course should use AI-related scenarios in exactly this way. The topic provides an opportunity to practise application, scepticism, ethics and board-level communication at the same time.
The annual report still needs someone behind it
AI is likely to become a normal part of corporate reporting.
That is not necessarily a problem.
Used properly, it can remove repetitive work, improve consistency and help reporting teams communicate complex information more clearly.
The danger appears when polished language begins to replace genuine thought.
An investor does not need to know whether every sentence was typed by a person.
They do need confidence that the conclusions behind those sentences belong to people who understand the business and are prepared to stand behind what has been reported.
That is the difference between automated writing and authentic reporting.
AI can generate the words.
Human judgement still has to give them meaning.
Comments