
Sign up to save your podcasts
Or


An OT patch is deferred because production cannot stop. The risk register says a firewall rule compensates for the exposure. But after maintenance, troubleshooting or a loss of connectivity with the SOC, what evidence shows that the rule still blocks the intended path?
This technical decision analysis uses a hypothetical plant scenario to compare configuration change alerts, baseline checks and tests of actual traffic reachability. It also examines what evidence the plant can preserve locally when central logging is unavailable, who owns escalation, and what should trigger a review of the patch deferral.
A valid diagnostic session can still issue a request the vehicle should refuse. This episode opens with a hypothetical remote-attack scenario, not a report of a documented incident. It asks when an ECU should reject a UDS RoutineControl (0x31) request because the vehicle state is unsafe, stale or cannot be verified.
We compare state checks at the gateway and at the target ECU, distinguish a diagnostic response from what the routine actually does, and discuss how hardware-in-the-loop fault injection can test those assumptions. The hardest question comes after the routine starts: what should happen if a required condition disappears during execution?
A shared maintenance password on a legacy train is a security risk. Replacing it can also affect diagnostic compatibility, supplier support and the evidence behind the Safety Case. What can the operator change now, and what needs a planned engineering change?
This technical decision analysis weighs individual access at the gateway, controls on maintenance sessions and a phased move away from shared credentials. It examines what those measures leave exposed at a local diagnostic port, what evidence the operator and supplier need before making changes, and how to compare the cost of keeping compensating controls with the cost of replacing the credential. The scenario illustrates an engineering decision; it is not a report of a documented incident.
A supplier declares an automotive ECU not affected by a vulnerability identified in a software component. What evidence should the manufacturer request before relying on that assessment?
This Friday episode explores a hypothetical supplier-assurance scenario involving closed-source firmware, a safety-related ECU and production constraints. The discussion contrasts component absence with code reachability, examines build-specific evidence and considers the trade-offs between accepting an assessment, requesting further investigation, applying containment and updating software.
Editorial clarification: this is not a documented vehicle incident. A VEX statement does not by itself determine a TARA risk rating or authorize risk retention. Binary analysis findings remain bounded by their method and coverage. The evidence package discussed is an assurance proposal, not a universal VEX requirement; regulatory and safety consequences require case-specific assessment.
More analysis: https://cybersecurityunderpressure.com
What if a PLC program change is flagged while a factory remains in production and process indicators appear normal? This Wednesday episode uses a hypothetical engineering scenario to examine the gap between apparent process stability and confidence in controller logic.
We compare possible change paths, engineering access, network visibility, controller state and the safety consequences of an unverified response. An online edit, a compromised engineering workstation and manipulation of operator views are hypotheses discussed in the dialogue, not established findings from a documented intrusion. The episode also examines the limits of live verification, the need for a trustworthy baseline and plant-safe recovery decisions.
This is an analytical scenario, not a report of a confirmed attack. Specific mechanisms and outcomes depend on the PLC model, configuration, tooling and evidence available at a real site.
A production line is stopped, and a supplier needs an engineering connection to diagnose the fault. In this explicitly hypothetical CUP scenario, the same legacy session can also download a PLC program. The question is who can constrain and end its authority when recovery is urgent.
We trace the path from supplier identity through remote access, a plant-controlled workstation, the engineering tool and the controller. A broker disconnect, account disable and controller-side termination do not necessarily have the same effect; none can undo a program already downloaded. MFA, recording and a jump host add control points, but do not make a change-capable tool read-only.
We discuss named access, expiry, local supervision, alternative routes, surviving sessions and a bounded non-production acceptance exercise. This is an engineering analysis, not a reported incident, product vulnerability or completed experiment. NIST SP 800-82 Rev. 3 provides context; the revocation sequence and acceptance checks are CUP recommendations.
Read the companion analysis: https://cybersecurityunderpressure.com/analysis-supplier-session-revocation-legacy-ot
What happens when a trusted note-taking workflow becomes a route into an industrial control environment? This episode examines an Evernote-to-SCADA attack path and the boundary between ordinary collaboration tools and operational access.
We consider how attackers can use familiar cloud content, links or shared documents to influence users and systems that support industrial operations. The practical questions are where trust is established, what privilege is needed to reach an engineering workstation or SCADA component, and which actions the available evidence actually supports. A suspicious document alone does not prove process manipulation.
For OT teams, the response starts with controlled access paths, isolation between corporate and plant networks, application and identity telemetry, and verification of changes at the engineering and control layers. Detection and containment must respect plant safety and availability. The lesson is to test the complete route from external content to privileged industrial action, not merely the first click.
An SBOM is only as useful as the components and relationships it captures. This episode examines how incomplete software bills of materials can hide critical vulnerabilities when transitive dependencies, embedded packages, firmware or build artifacts are missed.
We discuss the difference between generating an SBOM and accepting it as evidence: checking scope against the delivered build, reconciling package and version identifiers, testing dependency depth and validating the tool configuration. A clean vulnerability scan cannot establish absence of risk if the underlying inventory is incomplete.
For automotive and industrial systems, OEMs, suppliers and operators must trace findings to the exact delivered baseline, assess exploitability in context and maintain that evidence as products change. An SBOM supports this work; it does not replace product-level security analysis.
Application edit rights and operating-system administration are not the same security boundary. This CUP Micro Brief shows how an over-privileged video-management service can turn a limited application role into control of the underlying server.
Related CUP episodes:
Can a connected vehicle reject a forged emergency-braking warning without suppressing a legitimate safety message?
This episode examines false emergency electronic brake light warnings in connected-car and V2X environments. We break down the trust boundary between a valid sender credential, a plausible road event and the vehicle response that follows. Cryptographic validity can authenticate a source, but it does not automatically prove that the reported event is true.
The discussion covers message authentication, plausibility checks, certificate and credential revocation, duplicate and anomalous-event detection, safe degradation, incident response and the operational limits of automated blocking. It also separates prevention, detection and containment, because each control proves something different.
Key lessons:
• Bind message acceptance to authenticated and authorized senders
• Correlate warnings with vehicle, sensor and situational context
• Design revocation and trust updates for real driving conditions
• Avoid fail-open and fail-silent responses
• Validate the complete path from received warning to vehicle action
The episode does not assume that every V2X deployment is vulnerable or that a forged warning necessarily causes braking. Impact depends on the protocol, implementation, vehicle state and the control path exposed by the receiving system.
You can also listen to Cybersecurity Under Pressure on PlayPodcast.net.