On-Device Model Extraction / Theft

← AI/ML (OWASP LLM/GenAI Top 10)

ondevice_model_extraction   MEDIUM

Description

On-device model weights are readable/extractable from app storage, enabling model theft and offline attack crafting.

How it works

The on-device model is bundled in plaintext with no encryption or signature. Any attacker who can read the app package can copy the file and recover the full weights, and even an embedded system-prompt secret. The demo reads the bundled placeholder and shows its raw, fully extractable contents; the real theft is a simple file copy.

How to exercise it. DVMA is the harness - open this module from the home index and tap the demo action. The screen ships the malicious input and simulates the attacker (e.g. the companion app, crafted intent, or scanned payload) in-process, and the evidence panel prints the proof. The Tools (optional) and Attack inputs below are only needed to reproduce the exploit end-to-end on a real device.

Exploit steps

1. Set up. Build DVMA with a flavor that enables the AI/ML (OWASP LLM/GenAI Top 10) category (e.g. --dart-define-from-file=config/flavors/dev.json) and run on an emulator/simulator you control. The demo needs no external tooling; for the optional on-device reproduction the relevant tools are: objection, r2, frida, xxd · Android: adb, run-as · iOS: ifuse, afcclient.
2. Locate the target. From the home index, open On-Device Model Extraction / Theft (ondevice_model_extraction). The How it works section above describes this module’s specific weakness; the screen states the intended-secure behavior and exposes the vulnerable action.
3. Exploit. Send the crafted prompt / poisoned content to the in-app assistant and confirm the model obeys it - leaked system prompt/secret, an unconfirmed tool call, or attacker-controlled output.
4. Observe the evidence. Trigger the vulnerable action and read the evidence panel - it prints the concrete proof (leaked value, accepted replay, executed payload, or unauthorized result).
5. Contrast with the secure path. Run the module’s secure/hardened action (where provided) and confirm the same attack is rejected - this is what a correct implementation should do.
6. Map it back. Cross-reference the MASTG v2 test(s) MASTG-TEST-0302, MASTG-TEST-0300 for the canonical procedure and remediation.

Tools (optional)

• objection
• r2
• frida
• xxd

Android

• adb
• run-as

iOS

• ifuse
• afcclient