AI and ESG reporting

Tightening regulations, new voluntary frameworks and growing investor scrutiny are making sustainability reporting more demanding. AI could ease the load – but will it really transform the disclosure process? And can it be used ethically and sustainably?

An often-quoted figure among sustainability professionals is that environmental, social and governance (ESG) regulations have surged by 155% over the past decade.

As demand for rigorous and transparent disclosures grows — a shift largely welcomed by the industry — sustainability reporting is becoming more complex and time-intensive. Many ESG teams now find themselves consumed by compliance tasks, rather than strategy, innovation and long-term impact.

Current tools offer limited support. A recent report by software firm Workiva, based on a survey of 1,601 senior executives worldwide, found that 73% believe their existing reporting technology falls short of meeting the demands of new climate regulations.

Professionals are turning to AI to help automate routine tasks and make sense of sprawling data.

According to another Workiva survey of 2,300 corporate reporting experts across finance, sustainability, risk, legal and compliance, 88% have seen a rise in ROI from AI over the past year. Meanwhile, 96% reported saving time and 94% noted productivity improvements.

Mandi McReynolds, chief sustainability officer at Workiva, tells edie that her team has been beta testing AI tools in sustainability reporting for over two years.

“AI helps us cut through complexity,” McReynolds explains. “When working on the EU’s Corporate Sustainability Reporting Directive (CSRD), we noticed many companies struggle to identify which material issues are truly relevant out of the 50+ requirements. AI tools help narrow down that list and focus the team’s efforts.”

Beyond materiality assessments, McReynolds says, Workiva uses AI to align targets with global frameworks like the UN Sustainable Development Goals (SDGs), streamline compliance tasks, conduct peer benchmarking and draft sections of sustainability reports.

However, even as the benefits of the technology come into focus, many are taking a cautious approach.

Mastering prompts and measuring impact

Echoing McReynolds’ points, Deepa Rao, sustainability governance and climate change lead at Cognizant, tells edie that “prompt engineering is everything now”.

She explains that AI hallucinations often happen because users ask broad or unfocused questions, such as “Can you write the whole report for me?” Instead, breaking tasks into smaller, clear prompts reduces errors.

Rao stresses that AI does not replace human judgement: “It’ll do 50%” of the work, with the rest requiring careful review”.

While AI has improved efficiency, she notes that limitations remain, especially in design tasks like creating fully on-brand presentations, which still require human input. Overall, AI has helped her complete work faster and regain work-life balance.

She says: “I don’t eat lunch at my desk anymore. I enjoy it at my dining table. I don’t drink coffee at my desk either. That used to be because of the workload. Now, my work is done for the day and I’m already onto the next day’s tasks.”

As one of the few large firms to develop internal AI principles, edie asked Rao whether Cognizant is disclosing the carbon footprint of its AI use.

She says: “Not yet, but we want to. We’ve started considering it. I don’t know how we’ll do it, which is why I can’t tell you how — but we are looking into it.

“We’re trying to figure out how much AI we’re using and how much energy that’s consuming. I don’t know if anyone’s cracked that yet, but we’re working with SMEs to find out.

“But there are ways to offset it. For example, we travel less now, which helps compensate. It’s a trade-off.”

However, having governance principles in place has helped regardless.

“The big picture in this AI revolution is redefining roles and reassessing processes. Everything else follows,” Rao adds. “If you include AI in someone’s job description, if you define that they’re supposed to use AI for certain tasks, then you’ll see the efficiencies. If you don’t, you won’t.”

Cognizant has KPIs for its employees to measure how much resource and time can be saved using AI. Rao highlights that she has seen more than £100,000 in savings from improved resource and time efficiencies.

Building on the need for strong oversight, McReynolds of Workiva says companies should “clearly disclose their oversight structures and the framework they’re using to drive responsible AI practices”.

She emphasises that she is “not in favour of disclosure just for the sake of it”, adding that AI transparency should focus on material risks, opportunities, or costs.

On reporting AI’s carbon impact, McReynolds urges caution: “Before we rush to include AI’s carbon usage in emissions reports, let’s agree on standardised methods — whether through the GHG Protocol or another global framework — so that we can make meaningful, apples-to-apples comparisons.”