Attributes

Whole assembly · documented reference plus configured modules
Origin	Association	Properties
Fixed reference	On-board	PID-11
Fixed reference	On-board	PID-11
Fixed reference	On-board	PID-13
Fixed reference	On-board	PID-13
Fixed reference	On-board	PID-13
Fixed reference	On-board	PID-13
Fixed reference	On-board	PID-14
Fixed reference	On-board	PID-15

Two worked examples

PID-11 · Device includes a microprocessor

Side channels and fault injection

PID-11 maps to exactly four threats: TID-101 power analysis ↗, TID-102 electromagnetic analysis ↗, TID-103 microarchitectural side channels ↗ and TID-105 hardware fault injection – control flow modification ↗. TID-105 is where voltage, clock and electromagnetic fault concepts sit; EMB3D has no separate clock-injection threat. Its mitigations are MID-062 ↗ (leading), MID-063 ↗ and MID-066 ↗ (intermediate).

Fault injection appears here for research and learning. It is not a service LrrK offers by default or a capability this page demonstrates, and nothing here describes how to perform it. For reviewed public fault-injection evidence on drones and embedded devices, see TID-105 records in the Hardware Attack Library (HAL).

Load PID-11 in Attributes →Same selection at MITRE ↗

PID-1241 · Device includes DDR DRAM

Memory disturbance is a data threat

Selecting PID-1241 also includes its parents PID-124 and PID-12. PID-1241 itself maps to TID-110 ↗ (hardware fault injection – data manipulation), the DDR/Rowhammer case; its parent PID-124 adds TID-109 ↗. TID-110 covers data manipulation through DRAM disturbance such as Rowhammer. That is a different concern from physically induced control-flow faults (TID-105).

Load PID-1241 in Attributes →Same selection at MITRE ↗

Earlier LrrK HW labels

LrrK’s proposed HW-01–HW-09 labels are retired in favour of official EMB3D identifiers. Links to them land here.

HW-01 · Side-channel analysis
HW-02 · Fault injection
HW-03 · Malicious peripherals and device emulation
HW-04 · Supply-chain tampering and component substitution
HW-05 · Platform firmware and management-controller compromise
HW-06 · Memory disturbance: Rowhammer
HW-07 · Debug and service-interface abuse
HW-08 · Direct memory access abuse
HW-09 · Bus interception and interposer attacks

All 16 EMB3D hardware threats

The complete hardware threat list in MITRE EMB3D™ 2.0.2, for browsing. This list is independent of your selection above and is not a set of findings for any device. Each threat links to reviewed public evidence in the independent Hardware Attack Library (HAL).

Priority environments

Current engagements focus on consumer robotics. Industrial control, connected medical devices and AI compute are research and assessment priorities; suitability is confirmed during scope review. Physical testing requires separately agreed authority, access, capability, recovery and stop conditions.

Drones, robotics and autonomous systems: Can controller-firmware changes, credential extraction, maintenance-interface abuse or onboard-compute compromise affect movement or sensing?
ICS and PLCs: Who can change control logic, firmware or operating parameters? How is unauthorized change detected and safe restoration validated?
IoT and connected medical devices: Are updates authenticated, service interfaces protected and the support lifecycle understood? Is network access constrained?
AI accelerators and compute: Can firmware, BMCs, peripherals or memory alter execution despite an intact stored model? Which isolation assumptions hold?

Current offers and terms →

Evidence is not one category

Keep a researched mechanism, controlled research demonstration, observed operational incident and LrrK scoped finding separate. The cited work below is external evidence—not a LrrK finding or proof of operating effectiveness.

Observed operational attacks · July 30, 2026

Internet-facing water-sector PLCs

FBI/EPA reported remote compromise of Rockwell MicroLogix PLCs, configuration changes, reported project-file modification and operational disruption. This does not establish physical chip tampering.

FBI alert — primary incident report ↗

Research demonstrations · source releases cited; reference review November 1, 2026

GPU memory disturbance

GPUHammer demonstrated GDDR6 bit flips and model degradation on tested hardware. Subsequent GDDRHammer and GeForge research demonstrates page-table corruption and escalation under specified conditions. GeForge host-memory compromise requires IOMMU disabled. These results are not generalizable to every GPU or isolation setup; consult each release for its tested configuration and publication date.

GPUHammer ↗GDDRHammer / GeForge ↗

Controlled demonstration · October 29, 2026

AI infrastructure and runtime model behavior

Eclypsium’s controlled GH200/BlueField DMA proof of concept altered runtime model behavior while the stored artifact and prior integrity checks remained valid. Infrastructure access and configuration dependencies matter. This does not establish widespread exploitation.

Eclypsium / NVIDIA announcement ↗Primary technical report (PDF) ↗

Protect. Detect. Recover.

Firmware trust and measurement: Authenticated firmware updates, secure/measured boot and supported attestation. Validate what each measures or enforces; a TPM alone is not whole-device integrity. NIST SP 800-193 ↗
Inventory, baselines and recovery: Detect unauthorized changes and test trusted-image recovery as part of NIST’s protect/detect/recover model. Recovery must fit the actual component and operating constraints. Platform firmware resiliency ↗
Physical, service and peripheral boundaries: Control access and verify DMA isolation. Electrical fingerprinting is supplementary only where its accuracy has been validated. Kernel DMA Protection ↗ · OWASP service-interface reference ↗
Supply-chain custody: Approved suppliers, component identity, repair history and custody. Investigate substitutions and deviations without declaring all counterfeits malicious. NIST supply-chain guidance ↗
Operation and management networks: Remove internet exposure, isolate operation/management networks, broker remote access, limit peers and validate controller logic before restoring. FBI/EPA PLC advisory ↗

From a question to a handoff

The November 1 operating model, standard v0.1, is a proposed method—not a claim that its operating effectiveness is already proven.

Decision and bounded question
Name the deploy, update, operate or replace decision and the single question that matters to it.
Scope and owner authority
Declare hardware, firmware, configuration, time, sources and permitted access. Agree capability, recovery and stop conditions before testing.
Source or test → claim
Locate the evidence and method; keep researched mechanisms, demonstrations, operational incidents and scoped findings distinct. Separate observation from inference.
Independent human review
Challenge the claim, its support and its unknowns. No automatic security score substitutes for review.
Decision record and practical handoff
Deliver the reviewed conclusion and reusable evidence packet, with limitations, actions and reassessment triggers.
Observed outcome and learning
Where agreed and observed, record the outcome and what should change in the next question or review.

An Evidence Brief begins with one bounded question, an explicit source/method boundary, a reviewed conclusion and a reusable evidence packet. Assurance Desk is recurring human-governed scoped review or cadence after scope review—not automatic telemetry. A Product Security Passport is an evidence record, not certification; signing, publication and access depend on the agreed scope.

Evidence method →Current Desk scopes →

A finding carries its boundary

Use this guide to keep evidence inspectable and reusable. A citation or potential mechanism alone is not an observed finding. Record what was actually examined, what remains inferred and who independently reviewed the conclusion.

Affected hardware / firmware / configuration
Scope and time boundary
Required access and owner authority
Hypothesis / mechanism / MITRE EMB3D™ property (PID) and threat (TID) IDs (if applicable)
Method / source / evidence locator
Observed outcome
Inferred outcome and assumptions
Evidence status (researched mechanism / controlled demonstration / operational incident / scoped LrrK finding)
Mitigation / recovery / stop conditions
Residual risk / unknowns
Independent human reviewer
Reassessment trigger

Blank, generated locally; nothing is submitted or stored by this download. It is not an issued finding, Passport or certification.

Request an evidence brief →Assessment scope and request →

Source and license

Data: MITRE EMB3D™ site release 2.0.2, data model 2.0.1, from github.com/mitre/emb3d @ 0d7c25bb4e29 (2026-06-01). 59 properties, 81 threats, 89 mitigations, 343 model relationships. Bundled with this site; nothing is fetched from MITRE while you use it. LrrK is an independent implementation and is not endorsed by MITRE. EMB3D is not LrrK’s internal standard.

Source note: threats file lists PID-4114 for TID-330; PID-4114 is not among the 59 published properties and is omitted.

Full MITRE EMB3D™ license and disclaimers

LICENSE

The MITRE Corporation (MITRE) hereby grants you a non-exclusive, royalty-free license to use, copy, and create derivative works of EMB3D™ for internal business purposes or commercial use. Any copy you make for such purposes is authorized provided that you reproduce MITRE’s copyright designation and this license in any such copy and provide MITRE notice of such use at EMB3D@mitre.org.

For all other uses of EMB3D™, contact MITRE at techtransfer@mitre.org.

All rights not expressly granted are hereby reserved. Portions of EMB3D™ were produced for the U.S. Government under various contracts or funded by internal research and development funds and the U.S. Government retains certain rights in EMB3D™. You agree that you shall not include a charge for EMB3D™ in any sales or license of derivative products or services to the U.S. Government.

Your use of MITRE’s website and EMB3D™ will comply with all relevant laws, regulations, Executive Orders, and other obligations.

DISCLAIMERS

MITRE does not claim EMB3D™ enumerates all possibilities for the types of actions and behaviors documented as part of EMB3D™’s adversary model and framework of techniques. Using the information contained within EMB3D™ to address or cover full categories of techniques will not guarantee full defensive coverage as there may be undisclosed techniques or variations on existing techniques not documented by EMB3D™.

EMB3D™ AND ALL DOCUMENTS AND THE INFORMATION CONTAINED THEREIN ARE PROVIDED ON AN “AS IS” BASIS AND THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE MITRE CORPORATION, ITS BOARD OF TRUSTEES, OFFICERS, AGENTS, AND EMPLOYEES, DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, ANY WARRANTY THAT THE USE OF THE INFORMATION THEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF ACCURACY, NONINFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.

MITRE terms of use ↗Download license text

Components

Optional attachments