Who owns AI in your business?

Who owns AI in your business?

Four out of five food and agribusinesses are now using AI somewhere in their operation, yet 48% have nothing in place to govern it and only 11% have anybody responsible for it.

Adoption has moved much faster than management oversight.

The story first appeared in our 2026 Food & Agribusiness Report

Discover more insights like this in our 2026 Food & Agribusiness Report – Download Now!

View Report

Three years ago, this Report asked whether businesses were even aware of AI. At the time, 21% said they were using it. This year that figure is 80%.

The question now is whether businesses are managing its use effectively.

48% of respondents have nothing in place to support how their employees use AI. Almost nine in ten businesses (89%) do not have a designated person or team responsible for AI in the business and only 22% have a formal AI use policy.

80% are using AI

48% have nothing in place to support it

Food & Agribusiness Report 2026

Where AI is being used

Marketing, social media and content creation lead usage at 56%, followed by everyday productivity at 43%, market analysis and research at 39% and financial analysis and reporting at 33%.

Some of these uses involve sensitive or commercially important information. A third of respondents are using AI for sensitive tasks such as financial analysis and reporting, yet almost half of respondents have no rules covering which tools may be used, what data may be entered or how outputs should be checked.

Businesses are using more AI platforms

Last year the sector had a default, with ChatGPT at 67%. It is now 59%, while Copilot has climbed from 30% to 44%, helped by sitting inside Microsoft 365.

The growing number of platforms has a practical implication for AI policies. Avoid writing a policy around individual products, as the tools employees use will continue to change. Instead, set rules around the data such as what information can be entered into AI tools, what is off limits and what outputs require human review.

How regularly is AI displacing external or internal work?

45% of respondents have used AI in the last twelve months to do work they previously paid people to do. 38% have displaced external providers such as agencies, consultants, freelancers and contractors. 30% have displaced internal work, including roles they would otherwise have hired for.

Where AI is now performing work that was previously outsourced or carried out internally, businesses need to consider the operational dependency they are creating.

That dependency should form part of the business's wider risk management.

Additionally, when you stop using an agency or contractor, you also stop paying for the review step. Designers check artwork against the brand guidelines. Consultants check data and figures before they reach the board. Copywriters know what can and cannot be claimed on a label. All of that was bundled into the fee. If you move work from an external provider inhouse, you must intentionally build the review step into the process.

What is at risk?

Pricing, tender documents, supplier contracts, customer lists, recipes and formulations should not be entered into AI tools without understanding how that information will be handled.

Larger customers have started to ask about AI use in supplier questionnaires. If you xport, or supply a large retailer, expect the question and expect to need a written answer.

What is holding business back?

Lack of internal expertise, at 46%, and uncertainty about the relevant use cases, at 37%, are the two leading barriers. Data privacy and integration with existing systems follow at 33% and 32%. Cost of implementation sits near the bottom of the list at 9%.

Cost of implementation ranks well below expertise, use cases, privacy and integration as a barrier. The immediate challenge for most businesses is therefore less about budget and more about deciding where AI should be used, who is responsible for it and what controls are needed.

5 things to put in place

For businesses already using AI, five practical steps will strengthen oversight.

Name someone

Give one person responsibility. In a smaller business that is usually the operations manager or the financial controller. The role is to understand what staff are using, keep track of developments and provide a clear point of responsibility.

Ideally you should have an exchange rate mechanic built into all your pricing communication. Remember that exchange rates work both ways. It builds trust and shows fairness.

Keep the policy to one page

Approved tools, what must never go into them, what needs human review before it leaves the business and who to ask. A single page has a far better chance of being read.

Draw the data line clearly

Be specific about what is off limits, including customer and employee personal data, pricing and tender information, unpublished financials, formulations and product intellectual property and anything covered by a non-disclosure agreement. Where AI would help with that kind of work, pay for a business-tier account where your data is not used for training and carry out a risk assessment before uploading any information.

Be very wary of using free versions of any AI tool, especially where data is being uploaded.

Verify all outputs

A person checks anything with a number in it, any claim about a product, anything going on a label or pack and anything going to a customer, a bank or a regulator.

Train the people using it

Be very wary of using free versions of any AI tool, especially where data is being uploaded.

Faster than anything we have tracked

AI adoption has moved faster than any other technology trend tracked by this Report over the past nine years. Governance has not developed at the same pace. Businesses should continue to use AI where it creates value. However, as a business leader you also need to recognise that AI has to be managed with the same discipline applied to other important business systems and processes.

Talk to Amy Ward Whelan

Director of Food & Agribusiness1800 334 422 amywardwhelan@ifac.ie Linkedin