CyLab Security Academy

09/24/2026

AI security starter concept: A model can follow its rules correctly and still produce a dangerous result.

A perceptron classifies what it receives according to its inputs, weights, and bias. If an attacker manipulates the input, the model may place it on the wrong side of the decision boundary.

The calculation worked. The security failed.

Understanding that distinction prepares learners to examine more complex problems, including adversarial examples, distorted training data, and incomplete model assumptions.

The CyLab Security Academy connects AI fundamentals to the security judgment needed to evaluate how intelligent systems behave when conditions are no longer trustworthy.

Explore at learn.cylabacademy.org.



Carnegie Mellon University CyLab
Carnegie Mellon University

09/23/2026

AI starter concept: A perceptron does not understand the meaning of its decision.

During training, it compares its prediction with the expected answer. When the prediction is wrong, the model adjusts its weights. Repeated examples gradually shape the rule it uses to classify new inputs.

That distinction matters. A model can learn a useful mathematical pattern without understanding the real-world context surrounding it.

AI Foundations I gives learners a manageable model they can inspect step by step. That creates a foundation for asking harder questions about training data, model behavior, and security.

Learn more at learn.cylabacademy.org.



Carnegie Mellon University
Carnegie Mellon University CyLab

09/22/2026

AI starter concept: Inputs alone do not determine a model’s answer.

A perceptron assigns a weight to each input. Those weights determine which information has more influence. Its bias shifts the point at which the model changes from one classification to another.

In AI Foundations I, learners encounter variants with different perceptron parameters. They must determine how those changes affect the output instead of memorizing the result from one example.

Getting one answer right demonstrates completion. Explaining why the answer changes demonstrates understanding.

Explore AI learning at learn.cylabacademy.org.



Carnegie Mellon University
Carnegie Mellon University CyLab

09/21/2026

AI starter concept: What is a perceptron?

A perceptron is one of the simplest machine-learning models. It assigns a weight to each input, combines those values with a bias, and produces a classification.

The mathematics may be simple. The lesson is foundational: an AI system’s output is shaped by what it receives, how those inputs are weighted, and where its decision boundary is set.

The CyLab Security Academy’s AI Foundations I learning path starts with this visible decision process so learners can understand how a model reaches an answer before they begin asking whether that answer can be trusted.

Explore the Academy at learn.cylabacademy.org.


Carnegie Mellon University
Carnegie Mellon University CyLab

09/17/2026

A smart contract can execute its code exactly as written and still be insecure.

A reentrancy vulnerability can occur when a contract makes an external call before safely updating its own state. An attacker may be able to call the function again before the original transaction is complete.

Practice identifying this insecure coding pattern with Reentrance, a Blockchain challenge in the CyLab Security Academy Library.

Blockchain preserves what the code does. Security still depends on what the developer allowed the code to do.

Find more at cylabacademy.org.

Ripple

09/15/2026

Starter concept: A file can reveal information beyond what appears on the screen.

Metadata records details about a file, such as when it was created, what device or software produced it, or where it originated.

Practice finding that information with "information", an Easy Forensics challenge in the CyLab Security Academy Library.

The challenge introduces a core investigative habit: inspect the file itself, not only the content you can immediately see.

Find more at cylabacademy.org.

09/14/2026

Encoding 📣 changing a message's format so a machine or system can handle it. No secrecy involved. Anyone who knows the scheme can reverse it instantly. Think of writing a word in Morse code: dots and dashes instead of letters, but nothing is hidden. The "key" to Morse code is public knowledge.

Encryption 🔐 deliberately scrambling a message so only someone with a secret (a key) can read it. Without that key, the scrambled text should be practically impossible to unscramble.

Here's the test: if you can reverse it using only public, known rules, it was encoded. If you need a secret you don't have, it was encrypted.

Why this trips people up: encoded text often looks just as confusing as encrypted text. A string like `SGVsbG8gd29ybGQ=` looks cryptic, but it's just "Hello world" in Base64, a common encoding scheme with zero secrecy. Anyone can decode it in two seconds. Mistaking encoding for encryption is a classic beginner move, and it matters because encoded data offers no real protection at all.

So before you try to "crack" a suspicious string, ask: is this hidden, or just reformatted?

Explore hands-on learning at cylabacademy.org

A free cybersecurity education program and competition, created by security experts at Carnegie Mellon University.

09/10/2026

CTFs continue to be a prominent staple in the cybersecurity world for learning the skills that you need to succeed.
Shown here are the winners of MITRE's eCTF competition, who are all Carnegie Mellon students. Competitions like these continue to be a major source of knowledge and training grounds for improving your own skills and to stand out within a more competitive job world, whether it be with the Academy or MITRE.

08/28/2026

CyLab Security Academy is free for learners, but it is not free to build.

Challenges must be researched, developed, tested, and maintained. The platform must be ready when a teacher introduces a class or a learner decides to try cybersecurity for the first time.

Sponsors help sustain the infrastructure behind those moments.

This back-to-school season, help us keep hands-on cybersecurity and emerging-technology education accessible.

Learn more at cylabacademy.org.

08/27/2026

A learner’s opportunity to explore cybersecurity should not depend on whether their school can afford specialized software or build a technical lab.

CyLab Security Academy makes hands-on learning freely available, helping more students move from reading about technology to working directly with it.

Free access removes one barrier. Strong educational infrastructure removes another.

That infrastructure takes sustained support.