Concerns are mounting over the effectiveness of The Murray-Darling Basin Plan Review, with key stakeholders concerned about the use of artificial intelligence.
Questions have emerged around the use of artificial intelligence to assess public submissions, prompting leading advocacy body, the Murray Regional Strategy Group, to hand-deliver a letter of concern to Murray–Darling Basin Authority chief executive Andrew McConville.
MRSG chair Geoff Moar said the letter highlighted these concerns.
“Our members invest significant time, expertise, and resources into preparing evidence-based submissions, and we value a consultation process that reflects this effort,” Mr Moar said.
“We wish to raise a concern regarding the Terms and Conditions requiring submitters to agree to the use of artificial intelligence (AI) to evaluate and summarise submissions.”
He said submissions to the Basin Plan Review drew on complex hydrological, agricultural, economic and community‑level information, as well as personal experience, and stressed that such material required contextual understanding and subject‑matter expertise that AI tools could not reliably replicate.
“MRSG has already observed AI generated summaries in previous MDBA processes, specifically the 2025 Basin Plan Evaluation Questions, which contained factual errors and misinterpretations. This experience has reduced community confidence in AI-based assessment,” Mr Moar said.
The MRSG letter noted the Terms and Conditions did not explain what AI systems would be used, how they were trained, or how bias would be managed.
Mr Moar said without this transparency, submitters could not reasonably consent to AI-driven evaluation.
“MRSG is particularly concerned that AI models trained primarily on MDBA generated environmental material may default to an environmental first interpretation of evidence, unintentionally minimising or dismissing information relating to productivity, food and fibre output, or regional socioeconomic impacts.
“This creates a real risk that submissions offering balanced, practical, onground perspectives will be misrepresented or undervalued in the review process.”
Mr Moar emphasised that community confidence required a human-led review.
“Given the significance of the Basin Plan Review and the effort required to prepare submissions, MRSG expects that all submissions will be read and evaluated by qualified human reviewers,” Mr Moar said.
In its letter, the Murray Regional Strategy Group urges the Murray–Darling Basin Authority to guarantee that all submissions will be read and assessed by human experts, with any AI‑generated summaries checked by humans before they are used. The group also calls for full transparency around the AI systems and safeguards employed, and asks that submitters be allowed to opt out of AI‑based processing without penalty.
“These steps would help restore confidence in the integrity of the consultation process and ensure that Basin communities’ contributions are accurately understood and genuinely considered,” Mr Moar said.
