Study finds AI-generated sustainability reports were rated as more credible than real ones

A classroom exercise at the University of Auckland produced a troubling result: Even postgraduate students trained in accounting and sustainability were more likely to trust polished reports written by ChatGPT than real ones.

The finding suggests artificial intelligence could make deceptive marketing tactics like greenwashing more persuasive. 

What happened?

As Newsroom reported, an assignment devised by Ruth Dimes, then head of the University of Auckland Business School's business master's program, and accounting professor Charl de Villiers had postgraduate students use generative AI to write a CEO statement, a summary that would normally be included in a sustainability report, for a fictional company.

To test how the AI-generated statements would be received, the researchers mixed them with three authentic corporate reports and asked a separate group of students to evaluate the full set without being told which documents were real.

When students rated the reports for things like credibility and whether they would invest, the AI-produced versions came out ahead.

"They were very persuaded by the language in ChatGPT," Dimes said.

Despite the concerning results, the researchers noted the participants were not new to the topic.

"This was a really informed group of students," Dimes said, noting that they had already studied sustainability reporting and greenwashing.

What appealed to them, the researchers found, were reports that cast environmental and social results in a more favorable light than the underlying performance justified.

Why does it matter?

Corporate sustainability reports can shape investor confidence, public perception, and where money flows. If AI can produce reports that sound responsible while glossing over real environmental harm, it could become even harder to distinguish between companies making meaningful progress and those simply presenting a more polished narrative.

Newsroom cited Bloomberg data that estimates environmentally focused investments could exceed $40 trillion by 2030.

With major investments on the line, AI can be a powerful tool for greenwashing. In the experiment, researchers checked the AI-written statements with a tool designed to flag greenwashing. "And yes, they were quite greenwashy," de Villiers said.

He said that, unlike fabricated AI scenarios, actual company reports have to contend with real problems they cannot simply write off.

What are people saying?

According to Dimes, having students revise the chatbot-produced statements did not materially improve them.

"There needs to be some kind of control or regulation or guidance," she said, arguing that sustainability reporting still leaves too much room for selective wording and omission.

De Villiers pointed to another possible solution: investors asking for hard evidence instead of accepting polished promises at face value.

Otherwise, the risk is that shareholders and prospective investors may come to see sustainability reporting as little more than AI-generated fabrication.

Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.