It seems like AI is on everyone’s minds today. As you read articles about AI developments, you may encounter a range of complex terms used to discuss it.
This post will begin an “AI Glossary” series. In it, we’ll define key AI terms and explain the cybersecurity considerations related to them.
Agent: A standard AI chatbot just talks to you, but an agent can actually do things. A chatbot is like a consultant who gives you advice, whereas an agent is like an employee who takes that advice and files the paperwork or sends the email.
Algorithm: This is a set of rules a computer follows to solve a problem. Think of it like a recipe for a cake. If the computer follows the steps exactly, it gets the desired result every time.
API (Application Programming Interface): This is the bridge that lets two different software programs talk to each other. Think of it like a waiter in a restaurant. The kitchen (the software) and the customer (you) do not talk directly; the waiter (API) takes the order and brings the food back.
Cybersecurity considerations:
Some of the biggest risks associated with the use of AI occur when AI is in Agent mode. This is because the AI is enabled to undertake administrative actions such as making purchases and downloading files. This leads to risks such as the AI mistakenly downloading malicious files or making other decisions that compromise data.
AI should be used in Agent mode only with caution. If you do decide to use it, ensure that you have implemented the following:
-Limited permissions: Don’t allow it to access private data, bank account info, etc.
-Approval required to complete sensitive actions like purchases
If you need assistance with implementing these AI permissions, contact ASIS.