How Researchers Fully Cracked Kindle DRM Using Static and Dynamic Analysis to Extract the AES Key

Amazon Kindle protects millions of paid e‑books with a proprietary DRM system that derives encryption keys on the client using PBKDF2(accountSecrets + deviceID + credentialPath, iterations=100000). The key material consists of three components: Account Secrets stored in encrypted device storage, a unique Device ID, and a Credential Path pointing to the correct credential record.

Key Derivation Details

The researchers identified the exact PBKDF2 parameters used after 2023: 100,000 iterations, SHA‑256 as the hash algorithm, and a salt extracted from device‑specific data.

Static Analysis – Extracting Key Material

Using .NET decompilation of the Kindle for PC/Mac client, the team located the functions that handle key processing by searching for encryption‑related strings and call sites, then traced the data flow from user input to the final encrypted output.

Layered encryption architecture : an outer layer encrypted with a device key and an inner layer with an account key, meaning the device key alone cannot decrypt content.

Obfuscation techniques : critical functions are heavily obfuscated, complicating static analysis.

Dynamic Analysis – Capturing the Runtime Key

A full debugging environment was built using Frida. The following script attaches to Kindle.exe and intercepts the CryptoDecrypt export in the "AmazonKindleCore" module, logging the AES key when decryption is invoked:

# Using Frida for runtime tracing
import frida
session = frida.attach("Kindle.exe")
script = session.create_script("""
    Interceptor.attach(
        Module.findExportByName("AmazonKindleCore", "CryptoDecrypt"),
        {
            onEnter: function(args) {
                console.log("Decrypt called!");
                console.log("Key:", Hexdump(args[1], {length: 32}));
            }
        }
    );
""")

Three interception points were identified:

OpenSSL call used for AES decryption.

Pointer to the derived key material in memory.

Ion block parsing – Kindle stores metadata in the Ion binary format.

Ion blocks are structured as [4‑byte length][encrypted data][16‑byte IV][checksum], compressed with LZMA and encrypted with AES‑256‑CBC. After extracting the key, the researchers could fully decrypt and decompress the e‑book content.

Full Extraction Workflow

Prepare environment: install Kindle for PC and log in.

Inject Frida script into the Kindle process.

Open a protected e‑book to trigger decryption.

Capture the AES key during the CryptoDecrypt call.

Decrypt the e‑book using the captured key.

Decompress the LZMA payload to obtain the original text and images.

Tooling

The authors released a GitHub script that automates key location, processes multiple Ion blocks, and outputs the decrypted raw content. A simplified core function is:

# Core extraction logic (simplified)
def extract_aes_key():
    key_material = extract_from_memory()
    salt = get_device_salt()
    key = PBKDF2(key_material, salt, iterations=100000)
    return AES256Key(key)

Security Impact

The disclosed method shows a fundamental flaw: once the runtime key is extracted, the entire DRM protection collapses. While illegal for distributing copyrighted material, the technique can be used for personal backups, academic research, and security testing.

Legal and Ethical Considerations

Illicit uses: distributing protected content, bypassing copyright for commercial gain.

Legitimate uses: backing up purchased books, scholarly analysis, security assessments.

Amazon’s Potential Response

Update the DRM implementation with stronger runtime protections.

Change key‑derivation parameters.

Introduce additional anti‑debugging measures.

Defensive Recommendations for Content Providers

Use hardware‑based key storage (e.g., TEE).

Implement key rotation mechanisms.

Add runtime integrity checks.

Detect debugger attachment and monitor abnormal memory‑access patterns.

Identify known security‑research tools.

Conclusion

The study demonstrates that client‑side DRM can be systematically reverse‑engineered, emphasizing that technical protections must be complemented by legal measures to effectively safeguard copyrighted content.