Steganography

Hiding in Plain Sight

Encryption hides the content of a message. Steganography hides the existence of the message.

With encryption, Eve sees gibberish but knows you’re communicating secretly. With steganography, Eve sees a normal photo, text, or file and suspects nothing.

The goal isn’t “Eve can’t read the message.” It’s “Eve doesn’t know there’s a message at all.”

Encryption vs Steganography

You can combine both: encrypt the message, then hide it with steganography.

Terminology

Four key terms define the steganographic process:

• Cover medium: The innocent-looking carrier. A photo, text, audio file, video. Something that wouldn’t raise suspicion.

• Payload: The secret message you want to hide.

• Stego-object: The result after hiding the payload in the cover. Looks identical to the cover, but contains hidden data.

• Stego-key: Optional password or key needed to extract the payload.

Cover + Payload = Stego-object

Text-Based Steganography

First-Letter Method

Hide a message in the first letter of each word:

”Bring rope. I need gear.”

First letters spell: BRING

The sentence reads normally. The secret message hides in the structure.

Null Cipher

Hide the message at fixed positions. For example, every 5th word:

“The weather seems nice today but I think we should stay home and rest”

Every 5th word: today, we, rest

The text flows naturally. The hidden message is embedded in plain sight.

Text steganography exploits the flexibility of natural language. Many ways to say the same thing.

Image-Based Steganography: LSB

LSB stands for Least Significant Bit.

A pixel’s color is stored as numbers (0-255 for each of R, G, B). In binary, that’s 8 bits:

The last bit (highlighted) contributes almost nothing to the color. Changing it shifts the value by just ±1.

Invisible to the human eye.

The Technique

1. Take each pixel’s color value in binary
2. Replace the LSB with one bit of your secret message
3. Repeat for enough pixels to encode the full message

The color changed by 1 out of 255. Completely invisible.

Example: Hiding “Hi”

The ASCII for “Hi” in binary: 01001000 01101001

That’s 16 bits. We need to modify 16 pixels.

The image looks identical. The message is hidden inside.

Steganalysis

Steganalysis is the art of detecting hidden messages. The attacker’s countermeasure.

Common methods:

• Statistical analysis: LSB replacement creates detectable patterns in pixel distributions
• Visual inspection: Enhanced contrast might reveal anomalies
• File size analysis: Stego-objects might be larger than expected
• Comparison attacks: If Eve has the original cover, differences reveal everything

If steganography’s goal is undetectability, steganalysis tries to break that assumption.

Communication Channels

Not all hiding places are equal:

Unconstrained channels: You control the cover completely. You can choose any image, craft any text. Easier to hide data naturally.

Socially constrained channels: The cover must fit social context. Sending random nature photos to your boss would be suspicious. The cover must make sense for the relationship.

Good steganography considers social undetectability, not just technical.

A perfectly hidden message in a suspicious context still fails.

Applications

Key Insight

Encryption says “you can’t read this.”

Steganography says “there’s nothing here to read.”

The best secret is one nobody knows exists.