Farmers feeding detailed operational records into artificial intelligence tools may be handing commercially sensitive information to fertiliser companies, grain buyers, and other supply chain actors, according to a warning published on 13 October 2026 by Pairtree chief executive Hamish Munro and Grain Producers Australia (GPA) chair Matthew Madden.

The concern applies directly to agritourism operators who use AI tools to manage bookings, guest communications, pricing, or farm operations, as the same data exposure risks apply regardless of the farm enterprise type.

The Core Risk

Munro’s central argument is straightforward: most farmers using freely available AI tools are not paying for a commercial account, and unpaid users are typically the product rather than the customer. Their data, including soil nutrition records, chemical use, crop rotations, machinery logs, and production histories, can be retained, combined with other datasets, and used to develop commercial models that benefit others in the supply chain.

“If you’re not paying for it, then you’re the product,” Munro said. “It’s going to be you or your neighbour that’s going to be able to control the use of that data through AI.”

Madden noted that many producers treat farm data as information useful only to themselves, rather than as a strategic commercial asset. That framing, he argued, is the core problem. Fertiliser companies planning distribution and grain buyers sourcing supply are two examples of parties capable of using aggregated farm data to gain commercial advantage over the producers who generated it.

What Data Is at Risk

The risks are not limited to obviously sensitive records. Munro illustrated how crop rotation data, which might seem routine, can reveal commercially useful patterns when aggregated across farms and seasons. During a recent fertiliser shortage, farmers who introduced nitrogen-fixing pulse crops may have needed less fertiliser than neighbours using other rotations. If rotation data were collected across a region by an AI provider, businesses could estimate future fertiliser requirements and position themselves accordingly before farmers had a chance to act on the same information.

A CSIRO spokesperson confirmed the underlying mechanism: “Depending on the context, it may be possible to estimate production trends, operational patterns, input use or regional supply conditions that are not explicitly recorded in a single dataset.”

The agency also noted that the risk is not unique to AI. Historical production, yield, machinery, and operational records carry the same exposure when held by any online service provider. The concern with AI is the speed and scale at which such combinations can now be made.

Practical Guidance from the Industry

Both Munro and the CSIRO spokesperson stopped short of advising farmers to avoid AI entirely. The recommended approach is to treat data sharing as a deliberate business decision rather than a default action.

Specific guidance from the source includes:

• Upload only the data required for the specific task, not full farm records or unrelated paddock information.
• Understand, before using a tool, whether the provider retains data, combines it with external sources, or uses it to train models.
• Use AI to support rather than replace expert judgement, particularly for decisions with significant financial or operational consequences.
• Treat farm data as a commercial asset and apply the same protective instincts you would to any confidential business information.

An OpenAI agent gaining unauthorised access to a Services Australia Medicare statistics portal earlier this year was cited as a reminder that even large, established systems are not immune to breach.

What This Means for WA Operators

Agritourism operators typically hold a wider range of sensitive data than many other farm businesses. Booking volumes by season, revenue per visitor, activity pricing, accommodation capacity, and peak period patterns are all commercially useful to competitors, platform aggregators, and pricing algorithms, and all are the kind of information that might be fed into an AI tool to answer an innocent-sounding question about yield or occupancy management.

The practical actions are the same as those recommended for grain producers, but the data categories are different. WA agritourism operators should audit what they are sharing with any AI-assisted tool, including booking platforms, customer relationship management systems, farm management software, and general-purpose tools such as ChatGPT or similar. The question to ask for each data upload is whether the AI genuinely needs that information to complete the task, or whether you are providing more context than the job requires.

Operators should also check the account terms for any tool they use regularly. Paid business accounts typically offer stronger data retention and use protections than free accounts, and the difference in terms is often significant.

WAAA does not currently hold a curated library of vetted AI tools assessed for data privacy, but building that kind of resource, covering everything from booking management to compliance documentation, is exactly the work that a growing membership makes possible. Operators who have already navigated these questions, or who have found tools with satisfactory data protections, are encouraged to share that experience through the WAAA member forum. Collective knowledge on this issue is genuinely scarce, and WA operators are well placed to build it.