More greenwashing in sustainability reports

An experiment revealed that generative AI tools produce sustainability reports with more greenwashing, and participants did not mitigate this effect. In addition, the AI-generated reports were perceived as more positive and credible than the real ones, which suggests a risk of overreliance on the technology for environmental communication.

Over the last decade, the pressure to publish sustainability reports has led many organizations to try generative AI tools to draft their climate statements. The promise of speed and consistency looks attractive, but the technology can reproduce the same marketing tricks humans use to soften negative impacts. When the credibility of these documents becomes a strategic asset, any bias built into automatic generation turns into a reputational and regulatory risk that companies cannot afford to ignore.

To measure that risk, the authors of the paper published in *Business Strategy and the Environment* designed an experiment with 240 sustainability and corporate communications professionals. Each participant received a set of real environmental performance data and, using a generative AI tool, produced a sustainability report. An independent panel rated the level of greenwashing on a 0 to 10 index. The AI-generated reports averaged 7.3, against 4.1 for the manually written ones. Moreover, when asked to rate credibility and positive tone, 68% of participants rated the AI documents as more credible and optimistic than the real ones, even though the factual content was identical.

The finding forces leaders to rethink how they bring AI into their reporting processes. The evidence shows that automation without a layer of human review amplifies greenwashing and creates an overconfidence that regulators and critical stakeholders can exploit. Putting strategic controls in place, such as content audits, external verification metrics and multi-step approval protocols, becomes essential to preserve the integrity of environmental communication and avoid penalties or reputational damage.

Conclusions

• Generative AI can replicate corporate language to mask unsustainable practices, weakening real transparency.
• The perception that AI-generated reports are more credible points to a potential bias in how people evaluate automated content.
• It is crucial to put human and strategic controls in place so that AI does not distort sustainability communication.
• Business leaders and regulators must set up oversight frameworks so that AI does not replace ethical judgment in environmental reports.