The trust gap: why faster sustainability reporting needs stronger proof

AI is changing how sustainability information is produced, just as investors lean on it harder.

Using AI offers sustainability practitioners the immediate benefits of speed and efficiency. However, increased AI use is also making trust a bigger issue.

This article explores where the trust gap is opening, and how independent assurance can help you close it.

From disclosure to decisions

Sustainability data used to be something organisations reported. Increasingly, it is also something others act on. Investors, lenders, customers and boards use it to:

allocate capital;
price risk; and
choose partners.

The World Economic Forum (WEF)’s 2026 survey of Chief Sustainability Officers (CSOs) found that nearly two-thirds of respondents rated investor expectations as critical or very important.1 This placed investor expectations well ahead of those of customers and employees.

And when it comes to sustainability data, it's clear that, though the future of regulatory requirements may be harder to predict, investors' expectations have not softened.

This changes the question from “have we reported?” to “could we defend this number if an investor asked?”

The natural follow up is: who, other than us, has checked it?

How is the use of AI creating a 'trust gap' in sustainability reporting?

We all know that reporting is getting faster.

Boston Consulting Group (BCG)'s 2026 white paper ‘AI-Enabled ESG Reporting’ reports measurable gains from AI in:

data aggregation;
carbon accounting;
gap analysis; and
disclosure drafting.2

Over half of the CSOs in the WEF survey (52%) expect measurement and reporting to be among the areas where AI has the greatest positive impact. And finally, 57.8% of investors surveyed by Clarity AI already use AI in their analysis or plan to within a year.3

This is all good news for stretched reporting teams. But while AI promises efficiency, confidence has to be earned separately.

In the Clarity AI survey, investors using AI name accuracy as their leading concern about it (68.6%). It is reasonable to assume they are no more relaxed about the data they receive. CSOs raise a related issue: 48% say limited availability of reliable datasets is a major barrier to AI adoption.

None of this will surprise anyone who has been through assurance. Even mature programmes regularly turn up:

incomplete evidence trails;
inconsistent calculations; and
unclear ownership of data.

AI does not create these weaknesses, but it can repeat them faster, across more data points, with less human attention given to each one.

Consider a simple, hypothetical example:

Much real estate data comes from managing agents and other third parties, and a pipeline ingests the data as it arrives. So, suppose tenant-controlled areas for a group of assets were mapped into landlord-controlled energy data. Reviewed by hand, someone might question that jump. But in an automated pipeline, it would be applied to every affected asset, every month, and carried into next year’s baseline and target progress. Everything would be consistent, so nothing would look wrong.

This highlights a widening gap: organisations can now produce sustainability information faster than they can prove it is reliable.

Closing that gap is what independent assurance is for.

What are the challenges around AI-supported estimates for sustainability data?

According to BCG, in today's context, a single reported figure might combine:

metered readings;
supplier submissions; and
AI-supported estimates used to fill in gaps in value-chain data.

These estimates are legitimate and often unavoidable. However, a problem arises when a reader, or even the team signing off reported data, can no longer tell which parts were measured, which were estimated, and on what basis. A number that can't be explained can't really be defended.

The sources we have looked at agree with our view here at BIP.Verco: people remain accountable for the final number.

BCG argues that AI should be built into existing control and assurance frameworks, but that judgement, validation, and accountability should stay with people. In all three of its case studies, the final interpretation and sign-off of data stayed in human hands.

The WEF makes a similar point: technology rarely removes management challenges; it simply moves them.

BCG also suggests that oversight should scale with risk: the more AI shapes a conclusion that will be reported externally, the more clearly a person needs to own and document the decision.

This means that every person signing off on data needs to have the skills and objectivity to check what the AI did. And, of course, not everyone does. This opens up a whole new challenge. Consequently, as shown in the WEF survey, 68% of CSOs expect skills gaps to constrain how businesses use AI for sustainability.

Accountability needs evidence. When a model has done much of the work, the sign-off matters more, not less.

How can assurance help sustainability teams stay accountable in the age of AI?

BIP.Verco has an extensive track record of providing independent verification and assurance against standards such as ISO 14064-3, ISSA 5000 and AA1000AS. These standards assess whether reported information is:

accurate;
complete;
reliable; and
aligned with the chosen reporting criteria.

Assurance looks beyond the numbers to the data, methods, assumptions, systems, controls and governance behind them, so an automated pipeline is in scope like any other process.

Find out more about sustainability assurance

Here is what we test for an AI-enabled process:

It comes down to three questions (these apply whether a person or a model produced the figure):

Can it be traced?
Is there an evidence trail from source data, through each automated step, to the disclosed figure?
Can it be explained?
Are methods, assumptions, and estimates documented, applied consistently, and kept distinct from measured data?
Is someone accountable?
Are ownership, review points, and sign-off clear, and proportionate to how much the AI influences the outcome?

If you'd like to know how well you can answer these questions (and others), use our helpful, 12-question self-check document. It only takes about ten minutes and will help highlight where you can improve.

Simply fill out the form below to download and use the self-check document.

In summary: independent assurance can close the trust gap when AI is used in sustainability reporting

As BCG explains, AI can be used to check plausibility, flag anomalies and generate auditable exception reports. Used well, it can make a reporting process easier to assure. But then you need the assurance process to find out whether it has done so successfully.

Independent assurance:

assesses whether data sources are complete and reliable;
challenges if material automatic estimates are appropriate;
tests outputs through recalculation; and
evaluates whether human oversight and change controls operate effectively.

The assurance opinion is the visible output. For many organisations, the findings behind it are just as valuable: a clear view of where controls hold and where they don’t, before someone else (like an investor) finds out.

Where should you start with sustainability assurance?

The right starting point depends on where you are.

Not yet required to assure, or still developing your reporting processes?
We suggest a data and reporting methodology review. This assesses your data, methodologies and controls, showing what is working, where improvements are needed, and what to prioritise next.

About to go through assurance, including GRESB?
We suggest an assurance readiness review. This applies the assurance lens without a formal opinion and tests your position beforehand. Issues are brought to light while there is still time to fix them.

Ready to publish with confidence?
We suggest limited or reasonable assurance to ISSA 5000 or AA1000AS. This will give your stakeholders an independent opinion and your team practical findings.

Already assured for several years?
At this stage, you’ll want to reduce effort and costs each cycle, strengthening the controls that matter most. It’s worth checking that your process has kept pace with how your data is now produced.

AI will keep making sustainability reporting faster. The organisations that benefit most will be those that can prove speed has not come at the cost of accuracy.

We keep strict ethical walls between our assurance and advisory work, so independence is never in question.

If you're interested in taking the next steps on your assurance journey, please don't hesitate to get in touch.

Book a free chat with our assurance experts