Security testing
IoT penetration testing
An IoT penetration test attacks a connected device the way an attacker with one in hand would: through its hardware, its firmware, its radios, its companion app and the cloud behind it. It finds weaknesses a web pentest never reaches. It also runs into a problem web testers rarely face: working out what software is inside the firmware at all.
Updated
What an IoT pentest covers
| Layer | What testers attack |
|---|---|
| Hardware | Debug ports such as UART and JTAG, flash chips that can be read directly, tamper protection |
| Firmware | Extracted images: hard-coded credentials, keys, outdated components, insecure services |
| Boot and update | Secure boot, signature checks on updates, rollback to vulnerable versions |
| Radio and network | Bluetooth Low Energy, Wi-Fi, Zigbee, cellular and the local services the device exposes |
| Companion app | The mobile app that pairs with and controls the device |
| Cloud and APIs | The backend the device and app talk to, and whether one device can reach another's data |
How an IoT penetration test runs
- Scope and threat model. Who the attacker is: a remote attacker, a neighbor in radio range, or someone holding the device.
- Teardown. Opening the device, identifying chips and finding debug interfaces.
- Firmware extraction and analysis. Dumping the firmware, unpacking its file systems and working out what is inside.
- Interface testing. Attacking the radios, local services, app and cloud APIs.
- Exploitation and reporting. Proving impact and writing it up, as in any pentest report.
What firmware analysis cannot tell you
A tester working from a firmware image has to infer its contents. Binaries are stripped, versions are missing, components are compiled in statically, and vendor patches change code without changing version strings. Firmware analysis tools guess, and the list of components they produce is only as good as the guess. Days of the test go on this, and the result still misses things.
The device's build has none of that ambiguity. A Yocto or Buildroot build knows every recipe, version and source it fetched. Testers who receive that list start with facts, not guesses.
AI in IoT penetration testing
AI pentesting already helps with the parts of an IoT test that look like web testing: the companion app and the cloud APIs. AI tools also speed up firmware analysis by explaining disassembled code and spotting risky patterns. Hardware work, radio testing and judging the physical threat model still need a person with the device on the bench.
IoT pentesting and the Cyber Resilience Act
Connected devices sold in the EU fall under the Cyber Resilience Act. Its Annex I asks manufacturers to apply effective and regular tests and reviews of the security of the product, and to identify and document the components in it, including through an SBOM. A yearly IoT pentest covers part of the first. The second needs a component list for every firmware release, which the pentest does not provide. See penetration testing requirements for other frameworks.
Record the firmware build
CRACI does not run penetration tests. It runs your Yocto builds as GitHub Actions jobs on its own runners and records every source each build fetches: every component in Yocto's own SBOM, plus host packages, GitHub sources and Actions and other fetched sources, with a package URL for every Rust crate. Every firmware release gets an SBOM recorded from the build that produced it, the SBOMs are checked again as new vulnerabilities are published, and you can hand the same list to your pentester. See Yocto vulnerability scanning.
IoT penetration testing: frequently asked questions
What is IoT penetration testing?
An authorized attack on a connected device and its ecosystem: hardware, firmware, update mechanism, radios, companion app and cloud APIs, to find exploitable weaknesses.
How is IoT pentesting different from web pentesting?
Testers have physical access, so they attack hardware interfaces and extract firmware, and they test radio protocols. The app and cloud parts look like a normal web or API test.
What is firmware penetration testing?
Extracting a device's firmware and analyzing it for hard-coded secrets, outdated components and insecure services, then using what is found to attack the device.
Can AI do IoT penetration testing?
Partly. AI tools help with the app and cloud APIs and speed up firmware analysis. Hardware and radio testing still need a person with the device.
Does the Cyber Resilience Act require IoT penetration testing?
It requires effective and regular security testing and reviews, without naming the method, and an SBOM of the product's components. A pentest covers part of the first, not the second.
Give every firmware release a recorded SBOM
Run your Yocto builds on CRACI and record every component and source they fetch.
Book a demo