// research archive

Research

Case studies show the affected boundary, root cause, validation method, practical impact, disclosure status, and remediation. Working exploit material stays out of the public record until coordinated disclosure is complete.

Published research

Verified findings and public remediation.

01 / Moxa AIG-302 ThingsPro EdgePublished | CISA ICSA-26-293-03 | 2 CVEs

Secrets API authorization and embedded CA-key exposure in an industrial gateway

Two published vulnerabilities in Moxa AIG-302 firmware affected authorization for a secrets API and protection of a shared Root CA private key. Moxa and CISA credited the reporting researcher.

Read case study
02 / Lantronix G520Published | CISA ICSA-26-272-01 | 2 CVEs

Update integrity and package authenticity in an industrial gateway

Two vulnerabilities in the Lantronix G520 update and package-handling mechanisms could allow arbitrary code execution under the conditions described by CISA. The disclosure resulted in two CVEs, a public CISA ICS advisory, and vendor remediation.

Read case study
03 / MLflowPublished | CVE-2026-64849 | CVSS 3.1 9.3 Critical (CVE record)

Unauthenticated full-read SSRF in webhook delivery

A Critical SSRF flaw let an unauthenticated caller cross the MLflow server network boundary through webhook redirects and DNS rebinding.

Read case study
04 / Eclipse AnkaiosPublished | CVE-2026-84173 | CVSS 4.0 8.3 High

Authorization bypass in workload control rules

A flaw in the agent-side Control Interface authorizer allowed a scoped workload to access or modify cluster state outside its authorized subtree.

Read case study
05 / vLLMPublished | CVE-2026-73560 | CVSS 3.1 6.5 Medium

Multimodal media path bypassed hardened retrieval controls

A model-specific multimodal processor bypassed centralized media-retrieval protections, exposing network and local-file trust boundaries.

Read case study
06 / LMDeployPublished | CVE-2026-46517 | CVSS 3.1 7.8 High

Unsafe remote-code trust during model initialization

LMDeploy enabled Hugging Face remote code by default during model loading, removing the operator decision that peer inference systems expose as an explicit opt-in.

Read case study

Researcher-reported work

Clearly separated from public advisories.

These summaries identify work reported by the researcher when no matching public vendor acknowledgement or technical advisory could be verified. They do not claim a public CVE, vendor-confirmed impact, or independently verified upstream fix.

Google GenkitResearcher-reported

Provider configuration trust-boundary assessment

Researcher-reported work examined how application-controlled provider configuration crosses into an AI SDK transport boundary.

Read qualified summary

Research domains

Trust boundaries across connected systems.

01

AI Infrastructure

Model serving, inference systems, AI SDKs, agents, remote resources, model loading, cloud integrations, and isolation boundaries.

02

Robotics & Distributed Systems

Workload orchestration, distributed control, authorization boundaries, messaging infrastructure, robotics middleware, and component-to-component trust.

03

IoT & Industrial Systems

Embedded and connected devices, industrial software, firmware, device-management tools, network services, management interfaces, certificate validation, device communication, edge infrastructure, and supporting protocols.

04

Open Source Infrastructure

Security-sensitive libraries, container systems, parsers, protocols, package handling, cryptographic validation, and other upstream infrastructure.

IoT & Industrial Research

Active research

Current work examines security boundaries in connected and industrial systems, including management applications, firmware, embedded software, network-facing services, certificate validation, device protocols, and cloud-connected edge components.

Published industrial research includes 4 vulnerabilities across Moxa AIG-302 and Lantronix G520. CISA credits Ievgen Bondarenko in ICS Advisories ICSA-26-293-03 and ICSA-26-272-01, with vendor-fixed releases documented for both products.

Publicly documented findings are grouped in the published archive. Researcher-reported work without a matching public advisory is labelled separately, and findings still under vendor or coordinator review remain private until disclosure is complete.