CIS 580A4 · 3 credits · Spring 2027

AI-Driven Cybersecurity for Digital Business

Attackers started using AI. Defenders have to. In one semester you go from "I have never written code" to building spam filters, fraud detectors and anomaly detectors that work - then attacking them, so you understand how they break. No prior coding. No heavy math.

Already enrolled? Use Sign in. Want to take the course? Use Register in RAMweb. Just looking? Start with "What is this course?"

Signing in
About the course
Why it matters
What you build
Tools
Where it leads
How to register
Request access

Signing in takes about 30 seconds

The course website is private. Lecture slides, labs, quizzes and readings sit behind a sign-in, so only enrolled students can open them. There is no password to create or remember - you receive a six-digit code by email.

Open the course site. You will be sent to a Colorado State sign-in page.
Type the email address your instructor registered for you and choose Send login code.
Check that inbox for a six-digit code. Look in spam or junk if it is not there within a minute.
Enter the code. You stay signed in on that browser for 24 hours.

No code arrived?

The sign-in page always says "A code has been emailed to you", even when no code was actually sent. That is deliberate - it stops outsiders using the login form to discover who is enrolled. So that message is not proof one is on its way.

Your address is not registered

Access is granted per person. If your address has not been added, no code is sent. Email Dr. Abdunabi from the address you want to use and ask to be added.

You used a different address

A code only goes to the exact address on the roster. A @rams.colostate.edu alias is a different address from @colostate.edu. Say which one you want to use.

It went to spam

Check junk and spam before assuming anything is wrong, then use Resend on the sign-in page. Codes expire after 10 minutes.

About the course

A hands-on, intuition-first introduction to using AI to defend the digital business - and to defending AI itself. Built for business students, not engineers. Every week pairs an interactive lecture with a guided Python lab that runs in your browser.

No code neededStarter code is provided for every lab. You extend it rather than start from a blank page.

12 labsTwelve working things you built and can explain in an interview.

$0 in toolsEverything runs free in a browser. You need a laptop and a Google account.

Meets Monday & Wednesday, 10:00-11:15 am, Rockwell West (RWW) 118. Prerequisite: CIS 370 or CIS 606.

Why this matters right now

Security used to be a specialist's problem. It is now a business problem, because the tools on both sides changed at the same time. The same models that write your email also write phishing that does not look like phishing.

Bar chart: among breaches where attackers used AI, 37 percent involved AI-written phishing and 35 percent involved deepfake impersonation. — **One in six breaches now involves an attacker using AI.** The two most common uses are phishing written by a model and deepfake impersonation of someone you trust. IBM, *Cost of a Data Breach Report 2025*

The other half is newer, and almost nobody is trained for it: organizations now deploy AI of their own, and that AI is itself a target.

Bar chart: 97 percent of organizations breached through AI had no AI access controls, 63 percent have no AI governance policy, and 13 percent were attacked through their own AI models or applications. — **Almost every organization breached through its own AI had no access controls on it.** Most still have no governance policy at all. That gap is where this course points you. IBM, *Cost of a Data Breach Report 2025*

Bar chart: the global average data breach costs 4.44 million dollars, rising to 5.11 million where unapproved AI use is widespread. — **Employees quietly using unapproved AI tools add $670,000 to the average breach.** Not a hypothetical risk - a line item. IBM, *Cost of a Data Breach Report 2025*

You do not need to be technical to matter here. Most costly security failures are decisions, not exploits: who could access what, which tool someone pasted company data into, whether anyone checked the model before it shipped. This course is built for the people who will be in the room when those decisions get made.

What you will actually build

Twelve labs, each one a working thing you wrote and can explain.

A spam classifier

Train a model on real messages, measure how often it is wrong, and decide whether that is acceptable.

A phishing URL detector

Judge it the way a security team would: false positives cost money too.

A fraud & anomaly detector

Find the unusual login in a month of normal ones, without being told what unusual looks like.

An attack on your own model

Craft adversarial examples that fool a classifier you just built. Then defend it.

A prompt-injection break

Make an LLM tool ignore its instructions in a Capture the Flag challenge, then write the fix.

A live incident exercise

Work a CISA tabletop scenario in a team, deciding under pressure with incomplete information.

Tools you will use

All free, all in the browser. A laptop and a Google account is the whole list.

Python™kept at a gentle beginner level
Google Colabnothing to install
scikit-learnthe models you train
pandashandling security data
Matplotlibseeing what a model did
Jupyteroptional local alternative
H101 Hacker101guided hacking lessons
CTFd CTFdour in-class capture the flag
CISA CISA Tabletopsreal federal incident scenarios

Logos are the property of their respective owners and appear here only to identify the tools taught in this course. Hacker101, CTFd and CISA are shown as text: CISA's is a U.S. government seal, whose use is restricted because it can imply endorsement.

Where this leads

Bar chart: information security analyst jobs are projected to grow 21 percent from 2025 to 2035, against 3.5 percent for all occupations. — **Security analyst roles are projected to grow six times faster than the average job.** U.S. Bureau of Labor Statistics, *Occupational Outlook Handbook*

$129,180median pay for information security analysts, May 2025 BLS

+21%projected job growth 2025-2035, "much faster than average" BLS

Bachelor'stypical entry-level education - no graduate degree required BLS

This is not only a path into security roles. Every analyst, product manager and consultant now works alongside AI systems, and being the person who can ask whether a model is safe to ship is worth more than the job title suggests.

How to register

Registration happens in RAMweb, the university's own system - not on this page. You will need your CSU NetID.

Sign in to RAMweb with your NetID.
Open Registration, then Register for Classes, and choose the Spring 2027 term.
Search the schedule for CIS 580A4 - AI-Driven Cybersecurity for Digital Business - and add it to your cart.
Submit. The course appears in Canvas once registration is processed, and your access to the course website is added from the class roster.

Prerequisite: CIS 370 or CIS 606. If RAMweb blocks you for a prerequisite you have met another way, or the section shows as full, email Dr. Abdunabi - overrides are handled by the department, not by this page.

Requesting access

Access to the course website is managed by the instructor. To be added, email:

Ramadan Abdunabi, Ph.D.
ramadan.abdunabi@colostate.edu
Senior Clinical Professor, Computer Information Systems
017 Rockwell Hall, 1277 Campus Delivery, Fort Collins, CO 80523

Please include your full name, the course you are registered in, and the exact email address you want to sign in with. Access is usually added the same day during term.

Sources

Figures are current as of September 2026: IBM, Cost of a Data Breach Report 2025 (breach cost, attacker use of AI, AI access controls and governance) · U.S. Bureau of Labor Statistics, Occupational Outlook Handbook: Information Security Analysts (median pay May 2025, outlook 2025-35) · Employment Projections 2025-2035 (3.5% across all occupations).