Assessing AI risks before implementation

In this Help Net Security video, Frank Kim, SANS Institute Fellow, explains why more enterprises must consider many challenges before implementing advanced technology in their platforms.

Without adequately assessing and understanding the risks accompanying AI integration, organizations will not be able to harness its full potential or even just ensure that it’s in the best interest of their organization to utilize it.

Some important considerations to think about include:

Data risks – Ensuring that the data sent to GenAI models is handled securely and confidentially as not to raise privacy concerns and unintentional disclosure of private data.

LLM risks – The potential for malicious models, untrusted models, or other authorized model sharing could affect the quality of information put out by these GenAI models.

Application risks – Ensuring that access keys necessary for communication between various components in GenAI applications are safeguarded and kept from unauthorized exposure.