Skip to content
icon icon Building AI Intuition

Connecting the dots...

icon icon Building AI Intuition

Connecting the dots...

  • Home
  • ML Basics
  • Model Intuition
  • Encryption
  • Privacy Tech
  • Musings
  • About
  • Home
  • ML Basics
  • Model Intuition
  • Encryption
  • Privacy Tech
  • Musings
  • About
Close

Search

Subscribe
icon icon Building AI Intuition

Connecting the dots...

icon icon Building AI Intuition

Connecting the dots...

  • Home
  • ML Basics
  • Model Intuition
  • Encryption
  • Privacy Tech
  • Musings
  • About
  • Home
  • ML Basics
  • Model Intuition
  • Encryption
  • Privacy Tech
  • Musings
  • About
Close

Search

Subscribe
Recent Posts
March 1, 2026
Teaching AI Models: Gradient Descent
March 1, 2026
Needle in the Haystack: Embedding Training and Context Rot
March 1, 2026
Measuring Meaning: Cosine Similarity
February 28, 2026
AI Paradigm Shift: From Rules to Patterns
February 16, 2026
Seq2Seq Models: Basics behind LLMs
February 16, 2026
Word2Vec: Start of Dense Embeddings
February 13, 2026
Advertising in the Age of AI
February 8, 2026
Breaking the “Unbreakable” Encryption – Part 2
February 8, 2026
Breaking the “Unbreakable” Encryption – Part 1
February 8, 2026
ML Foundations – Linear Combinations to Logistic Regression
February 2, 2026
Privacy Enhancing Technologies – Introduction
February 2, 2026
Privacy Enhancing Technologies (PETs) — Part 3
February 2, 2026
Privacy Enhancing Technologies (PETs) — Part 2
February 2, 2026
Privacy Enhancing Technologies (PETs) — Part 1
February 2, 2026
An Intuitive Guide to CNNs and RNNs
February 2, 2026
Making Sense Of Embeddings
November 9, 2025
How CNNs Actually Work
August 17, 2025
How Smart Vector Search Works
Privacy Tech

Privacy Enhancing Technologies (PETs) — Part 2

Secure Collaboration Without Sharing Raw Data In Part 1, we covered how individual organizations protect data…

Musings

Advertising in the Age of AI

When you search for a product today, ads quietly shape what you notice. When you scroll Instagram, ads compete for…

Machine Learning Basics

Measuring Meaning: Cosine Similarity

Post 2b/N In the previous posts, we established that embeddings turn everything into points in space and that Word2Vec…

Machine Learning Basics

Making Sense Of Embeddings

Post 2/N When you search on Amazon for “running shoes,” the system doesn’t just look for those exact…

Machine Learning Basics Model Intuition

Teaching AI Models: Gradient Descent

Post 1b/N In the last post, we established the big idea: machine learning is about finding patterns from data instead…

Machine Learning Basics

Breaking the “Unbreakable” Encryption – Part 1

If you’ve spent any time in tech, you’ve heard of AES, RSA, and Diffie-Hellman. We treat them like digital…

Home/Machine Learning Basics/Breaking the “Unbreakable” Encryption – Part 1
Machine Learning Basics

Breaking the “Unbreakable” Encryption – Part 1

By Archit Sharma
6 Min Read
0
Updated on February 28, 2026

If you’ve spent any time in tech, you’ve heard of AES, RSA, and Diffie-Hellman. We treat them like digital duct tape—they just work, they keep our data safe, and we don’t think about them.

But there’s a massive “Quantum Shadow” looming over our current security. Some of our locks will hold up just fine, while others are about to be sliced through like butter.

This post breaks down how we lock data today, why some of it is at risk, and the weird “magic” of how we share secrets in public.


The Two Ways to Lock a Door: Symmetric vs. Asymmetric

Think of encryption as two different types of physical security.

The One-Sentence Definition

The property that defines whether encryption is symmetric or asymmetric is Knowledge Parity:

  • Symmetric: Both parties know the exact same secret. (If I know it, you know it).
  • Asymmetric: There is a knowledge gap. I know something (Private Key) that the world cannot know, even though they have the “other half” (Public Key).
1. Symmetric Encryption (The Safe)

This is the oldest trick in the book. You have one key. It locks the box, and that same key unlocks it.

Mental Model: The High-Security Safe If I want to send you a secret, I put it in a safe, lock it with my key, and send the safe to you. You use your identical copy of the key to open it.

The Catch: How did you get that key? If I mailed it to you and a hacker intercepted the mail, they now have total access to everything I ever send you.

2. Asymmetric Encryption (The Mailbox)

This is the “magic” that makes the internet work. Instead of one key, you have a Public Key and a Private Key.

Mental Model: The Open Mailbox Think of your Public Key as an open mailbox sitting on your porch. Anyone can walk up and drop a letter in it. But once it’s inside, only you (the owner with the Private Key) can open the back of the mailbox to read the letters.

The Strength: I don’t need to meet you in an alleyway to exchange a secret key. I just look up your public “mailbox” and drop my message in.


Asymmetric Part 2: Digital Signatures (The ID Card)

We talked about using a Public Key to lock a message. But if you flip the process, you get a Digital Signature.

  • The Concept: I “encrypt” a piece of data with my Private Key.
  • The Result: Since only I have that key, anyone with my Public Key can “unlock” it. If it unlocks successfully, they know for a fact it came from me.
  • The Use Case (Certificates): When you visit google.com, Google shows you a “Certificate.” This is basically a digital ID card signed by a trusted third party (like DigiCert). They used their Private Key to sign it, and your browser uses their Public Key to verify it.

Plaintext

Encryption:  Public Key locks  → Private Key unlocks
             (Anyone can send you a secret)

Signature:   Private Key signs → Public Key verifies
             (Only you can prove it came from you)




The Hybrid Reality: Best of Both Worlds

In the real world (like when you visit your bank’s website), we use a combination:

  1. We use Asymmetric math to “meet” and agree on a secret.
  2. Once we have that secret, we switch to Symmetric encryption (like AES-256) because it’s way faster and incredibly strong.

This is how HTTPS works. The asymmetric handshake establishes trust and exchanges a key. Then AES takes over for the actual data transfer.


The Magic of Diffie-Hellman: Sharing a Secret in Public

The most mind-blowing part of cryptography is how two people can create the exact same secret key without ever actually sending that key to each other.

Mental Model: The Paint Mixing Trick

Imagine Alice and Bob want to agree on a secret color, but a hacker is watching everything they do.

  1. Step 1: They agree on a common “base” color (Yellow) in public.
  2. Step 2: Alice picks a secret color (Red) and mixes it with Yellow → Orange. She sends Orange to Bob.
  3. Step 3: Bob picks a secret color (Blue) and mixes it with Yellow → Green. He sends Green to Alice.
  4. Step 4: The Magic —
    • Alice takes Bob’s Green and adds her secret Red.
    • Bob takes Alice’s Orange and adds his secret Blue.

Result: Both end up with Yellow + Red + Blue = same “Muddy Brown.”

A hacker watching the whole time saw the Yellow, the Orange, and the Green—but they can’t “un-mix” the colors to figure out the secret Red or Blue. They are stuck without the final key.

Why It Actually Works: The Commutative Property

In math, 2×3 is the same as 3×2. Order doesn’t matter. In Diffie-Hellman, Alice has her secret (a) and Bob has his secret (b):

  • Alice calculates: ga, then Bob adds b→(ga)b
  • Bob calculates: gb, then Alice adds a→(gb)a
  • Both equal: gab

Because the math allows secrets to be applied in any order and still reach the same destination, they “meet” at the same number without ever revealing their secrets to the hacker watching the wire.


The 2×2 Security Matrix: Where the Quantum Shadow Falls

This grid is the best way to visualize where quantum computers actually threaten us. We have two dimensions: how we get the key and how we use the key.

Symmetric Key Agreement (Manual/Pre-shared)Asymmetric Key Agreement (DH/RSA)
Symmetric Encryption (AES)BOX 1: THE BUNKER
Example: BitLocker, Corporate VPN
Status: ✅ SAFE
BOX 2: THE MODERN WEB
Example: HTTPS, WhatsApp, Zoom
Status: ❌ DANGEROUS
Asymmetric Encryption (RSA/ECC)BOX 3: THE HYBRID SIGNER
Example: Local Code Signing
Status: ❌ DANGEROUS
BOX 4: THE ULTIMATE RISK
Example: Pure PGP Email, Certificates
Status: ❌ CRITICAL

Export to Sheets

Why Box 2 Is the Most Common (and Most Dangerous)

Most of the internet lives in Box 2. We use Asymmetric math (Diffie-Hellman) to agree on a key, then Symmetric math (AES) to talk.

The Problem: Even though the “Talking” part (AES) is safe, the “Meeting” part (DH) is broken by quantum algorithms. If the hacker breaks the meeting, they steal the key for the talk. It’s like having a diamond-encrusted vault but sending the key to your friend via carrier pigeon. If a quantum “sniper” hits the pigeon, the vault doesn’t matter.

Why Box 1 Is the “Bunker”

In Box 1, you and I meet in person and I hand you a USB stick with a random 256-bit key.

  • There is no math for a quantum computer to solve.
  • There is no “period” or “rhythm” to find.
  • The only way a quantum computer wins is by guessing, and AES-256 makes that haystack impossibly large.

Why Quantum Computing Changes Everything

Quantum computers aren’t just “faster” computers—they think differently. Two algorithms pose the threat:

Shor’s Algorithm: The Asymmetric Killer

Asymmetric math (RSA, Diffie-Hellman, ECC) relies on “one-way” math—it’s easy to multiply two giant prime numbers, but nearly impossible to work backward.

Shor’s Algorithm is like a “Math Cheat Code.” It finds the “rhythm” or period of the math. Once a quantum computer finds that beat, the one-way math collapses. It can “un-mix” the paint.

Grover’s Algorithm: The Symmetric Nuisance

Symmetric encryption (AES) doesn’t have a rhythm to find. It’s just a chaotic shuffle. Grover’s Algorithm is like a high-speed search engine. It makes guessing the key faster—but doesn’t break the math itself.

The Result:

  • AES-128 with Grover’s → feels like AES-64 → HACKABLE
  • AES-256 with Grover’s → feels like AES-128 → STILL SAFE
The “Store Now, Decrypt Later” Threat

Here’s why this matters today, not just in some distant quantum future: A hacker can record your encrypted traffic right now. It looks like gibberish today. But in 10-15 years, when quantum computers mature, they can decrypt everything they stored.

If your data has long-term value (medical records, government secrets, financial data), the quantum threat is already here.


Summary: The Crisis in One Sentence

  • Asymmetric = A public lock and a private key. (Vulnerable to Shor’s “Math Shortcut”)
  • Symmetric = A shared secret. (Vulnerable only to Grover’s “Fast Guessing”)

The Crisis: We rely on the VULNERABLE one to deliver the SAFE one.


Final Thought

Encryption isn’t about one single lock—it’s about the entire chain of how we share and store secrets.

  • The symmetric safe (AES-256) is quantum-resistant. Just use bigger keys.
  • The asymmetric courier (Diffie-Hellman, RSA) is quantum-vulnerable.

A hacker can “listen” to the exchange today and decrypt it years from now. This is why the industry is racing toward Post-Quantum Cryptography—replacing the “clock math” of Diffie-Hellman with “Lattice math” that even Shor’s Algorithm can’t crack.

In Part 2, we’ll go deeper into how Diffie-Hellman, Shor’s, and Grover’s actually work—and why Bitcoin is sitting right in the middle of the quantum crosshairs.

Related Posts:

  • Breaking the "Unbreakable" Encryption - Part 2
  • How CNNs Actually Work
  • Needle in the Haystack: Embedding Training and Context Rot
  • Privacy Enhancing Technologies - Introduction
  • Privacy Enhancing Technologies (PETs) — Part 2
  • Teaching AI Models: Gradient Descent

Tags:

asymmetric-encryptioncryptographycybersecurityEncryptionquantum-computingsecuritysymmetric-encryptiontechnology
Author

Archit Sharma

Follow Me
Other Articles
Previous

ML Foundations – Linear Combinations to Logistic Regression

Next

Breaking the “Unbreakable” Encryption – Part 2

No Comment! Be the first one.

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Categories

icons8 pencil 100
ML Basics

Back to the basics

screenshot 1
Model Intuition

Build model intuition

icons8 lock 100 (1)
Encryption

How encryption works

icons8 gears 100
Privacy Tech

What protects privacy

screenshot 4
Musings

Writing is thinking

Recent Posts

  • Teaching AI Models: Gradient Descent
  • Needle in the Haystack: Embedding Training and Context Rot
  • Measuring Meaning: Cosine Similarity
  • AI Paradigm Shift: From Rules to Patterns
  • Seq2Seq Models: Basics behind LLMs
  • Word2Vec: Start of Dense Embeddings
  • Advertising in the Age of AI
  • Breaking the “Unbreakable” Encryption – Part 2
  • Breaking the “Unbreakable” Encryption – Part 1
  • ML Foundations – Linear Combinations to Logistic Regression
Copyright 2026 — Building AI Intuition. All rights reserved. Blogsy WordPress Theme