DEF CON 32 - Nano Enigma Uncovering the Secrets in eFuse Memories - Michal Grygarek, Martin Petr

The video demonstrates how euse-based memory in ESP32 chips can be physically extracted and read using affordable, accessible tools, revealing secure boot keys and exposing flaws in traditional physical protection methods like layering. This proves that such security measures are easily bypassed with basic equipment.

Key Takeaways:
• Euse-based memory stores data permanently and can be physically read via microscopy after decapsulation 12:13.
• Chips can be decapsulated using chemical methods or simple tools like red stone and diamond lapping, making attacks feasible at home 17:00.
• Scanning electron microscopy (SEM) enables high-resolution imaging to detect burned (1) and unburned (0) euses, with data extracted manually or via machine learning 14:00.
• A machine learning model achieves 99% detection accuracy by analyzing fused patterns in high-res images, automating the extraction of 64x64 binary arrays 27:00.
• The extracted flash encryption keys allow decryption of firmware, revealing hidden data like "hello Las Vegas" 34:00.

Euse-based memory is inherently insecure due to physical accessibility, and traditional protections like layering are ineffective against determined attackers with basic tools.

Sources:

• 12:13 Explanation of euse memory and its irreversible bit programming
• 17:00 Decapsulation using chemical and mechanical methods
• 14:00 Use of SEM for detecting burned/unburned euses
• 27:00 Machine learning model for automated euse detection an