________________________________________________________________________________________________________________________
5. Security is a common concern when it comes to using AI LLMs like ChatGPT – how does one tackle this when the food R & D is incredibly IP-sensitive? Formulations and process knowledge are highly sensitive IP, so companies should always start by checking whether any AI platform they use is SOC 2 compliant and built to enterprise security standards. Beyond that, it’ s important to understand how data is isolated and where it lives. For example, platforms can be deployed in private cloud environments, such as tenantisolated deployments on Google Cloud Platform, offering bank-level security even at an entry level.
For organizations with the highest security requirements, it’ s also possible to run fully private or on-premise deployments, including air-gapped systems where models operate entirely inside a company’ s own infrastructure. While those approaches are more complex and costly, they give organizations maximum control over their data.
The key point is that AI can be adopted securely in food R & D, but only if the underlying architecture is designed for privacy, isolation, and governance from the outset.
6. Now that we’ ve established the benefits and hurdles of AI, are there any barriers that exist before wider AI integration is a reality within food companies? The major barrier we see is people’ s openness and comfort level with using AI tools in a way that is productive for the company. People are growing more comfortable with AI, and we are seeing a shift within organizations to leverage it. For example, we’ ve seen an increase in job roles on digital transformation, even within
14