# UN Regulation No. 155 and ISO/SAE 21434:2021 > A road vehicle with a learned component is already inside an audited cybersecurity regime. UN R155 requires an approved Cyber Security Management System (CSMS), whose certificate is valid for three years at most, and ISO/SAE 21434 is the process framework those audits lean on. Provael supplies adversarial-robustness measurements on a learned policy that a CSMS can hold as one input among many. Regulation/standard: UN Regulation No. 155 · ISO/SAE 21434:2021 Timing: Applies in the EU to new vehicle types since 6 July 2022 and to all new registrations since 7 July 2024; the CSMS certificate is re-audited on a cycle. ## From the regulatory clock (as of the clock's last verification) - **UN Regulation No. 155, uniform provisions concerning the approval of vehicles with regards to cybersecurity and cybersecurity management system** - Binding law (for the actor it names). Applies since 6 July 2022 — in application as of 19 September 2026. Responsible: Vehicle manufacturer holding the CSMS certificate and filing the vehicle-type evidence with the Approval Authority. Provael measures toward it: Adversarial-robustness measurements on a learned policy, held by a Cyber Security Management System as one input among many: the likelihood column for the learned component's rows of the para. 7.3.3 risk assessment, one test in the para. 7.3.6 set, one record of a para. 7.2.2.2(e) testing process. It does not establish: No type approval, no CSMS certificate, no conformity position under R155 or ISO/SAE 21434; not legal advice. Type approval is granted by a type-approval authority against a CSMS. Simulation only; zero real-hardware results. Source: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:42021X0387 (verified 19 September 2026) ## What a Provael report can be used for - **R155 para. 7.3.3** - An exhaustive risk assessment for the vehicle type, considering the threats in Annex 5 Part A. Annex 5 names manipulation of vehicle parameters and malicious messages; a learned policy that changes behaviour when an instruction is reframed is a threat the table does not yet name, which is why its likelihood has to be measured rather than looked up. - **R155 para. 7.3.6** - Appropriate and sufficient testing to verify the effectiveness of the security measures implemented, before approval. A measured attack-success rate with a benign control is one test in that set. Whether the set is appropriate and sufficient is the Approval Authority’s judgement. - **R155 para. 7.2.2.2(e)** - The processes used for testing the cybersecurity of a vehicle type. A seeded, resumable run whose report is bound to its execution manifest by digest is one record such a process produces, and one an assessor can re-run. - **ISO/SAE 21434 Clause 15** - Threat analysis and risk assessment methods. The attack-feasibility and impact judgements for the learned component’s threat scenarios can carry a denominator instead of a qualitative likelihood. - A measured envelope-exit rate per risk with its 95% Wilson interval and its benign-FPR control, stated under the run's predicate, as the likelihood column for the learned component’s rows of the R155 risk assessment and the 21434 TARA - a simulation rate, not an in-service incident likelihood. - A seeded, resumable run bound to its execution manifest by digest, with the attack catalogue it ran, as one record of a CSMS testing process. - Verification evidence for a cybersecurity requirement placed on the learned component under Clause 10, re-run in CI on every retrain: a rate measured once is a rate about a model you have already replaced. - Floor results in the same artifact: attack families that were tested and did not separate from the benign control are reported at their own n, so the record does not contain only the risks that happened to fire. It also maps to: - **UN R156** - The software update and software update management system regulation applied on the same schedule. Every retrain of a deployed policy is a software update inside that system; a rate that is only measured once does not follow the model through it. - **Not mapped: ISO/SAE 21434 Clause 11, R155 paras. 7.3.4, 7.3.7, 7.3.8** - Clause 11 validates an item at the vehicle level, and a simulated result about one component does not reach it. The R155 paragraphs on implementing mitigations, on detection and forensics on the vehicle, and on cryptography have nothing a red-team run is on-point for. Leaving them out is what keeps the four rows above from reading as coverage of the whole specification. ## What it cannot be used for A mapping identifies possible relevance. It does not establish that this framework applies to your product, that anything here is compliance, that an assessor or insurer will accept it, or a real-world risk probability. Whether it applies to your product depends on your role, market and product class: a question for your counsel, not a determination this page makes. - The physical-transfer study is pre-registered and has not been run, and no independent party has reproduced the published run. - The keep-out predicate is uncalibrated. A success means the policy left its benign envelope, not that a danger threshold anyone has set was crossed. The measured headline result stands on one policy and one suite. - If a supplier tells you a simulation result satisfies R155, that is the claim to push on, including when the supplier is us. - Simulation evidence for the tested checkpoint and protocol. Physical safety and conformity require separate assessment. - Provael measures adversarial robustness - susceptibility to manipulation - not general accuracy, reliability, or functional safety. - The measured result uses templated, auditable attacks; the search-based families (optimized, universal_patch, gradient_patch) have no real-policy rate. Results are a floor on susceptibility - a behavioural lower bound, not a certified worst-case bound. **Evidence, not certification.** Running Provael does not make a system compliant or certified - it generates measurements you can put into a conformity or assurance file. Provael is not a notified body, a lab, or a certification scheme. Independent project. Not affiliated with or endorsed by ISO, the EU, NIST, IEC, OWASP, or MITRE. Not legal advice. Clause references are indicative: check each against the text you hold before you cite it. ## What it is - UN Regulation No. 155 is a UNECE regulation applied through type approval. Its paragraph 1 applies it to vehicles of categories M and N, to category O where fitted with at least one electronic control unit, and to L6 and L7 where equipped with automated driving functionalities from level 3 onwards. A warehouse AGV that is machinery rather than a road vehicle sits under the Machinery Regulation instead, and which of the two binds a given machine is a question for regulatory counsel. - It has two halves. Paragraph 7.2 is the Cyber Security Management System: the manufacturer must demonstrate the processes it uses to manage cybersecurity, identify and assess risks to vehicle types, verify that they are managed, test the cybersecurity of a vehicle type, keep the assessment current, and monitor for and respond to attacks. Paragraph 7.3 is the vehicle type: an exhaustive risk assessment against the threats in Annex 5 Part A, proportionate mitigations, and appropriate and sufficient testing before approval. - The audit is a cycle. Paragraph 6.7: the Certificate of Compliance for CSMS remains valid for a maximum of three years from the date of deliverance unless it is withdrawn, and paragraph 6.8 lets the Approval Authority verify at any time that the requirements continue to be met. There is no once-and-done approval to point a result at. - ISO/SAE 21434:2021, Edition 1, was prepared by ISO/TC 22/SC 32 with SAE and supersedes SAE J3061:2016. It specifies engineering requirements for cybersecurity risk management across the lifecycle of electrical and electronic systems in road vehicles, prescribes no specific technology or solution, and states that it can be used to implement a cybersecurity management system. Clause 9 is the concept phase, Clause 10 product development, Clause 11 cybersecurity validation at the vehicle level, and Clause 15 the threat analysis and risk assessment methods. ## Dates (clock entry last verified 19 September 2026; each line shows when it was checked) - UN R155 in force: 22 January 2021 (Read from the Regulation as published in OJ L 82, 9.3.2021 (CELEX 42021X0387) (checked 18 September 2026)) - EU, new vehicle types: 6 July 2022 (Regulation (EU) 2019/2144, Annex II row D4 (protection of vehicle against cyberattacks, UN Regulation No 155), date for refusal to grant EU type-approval; read from the consolidated text of 7 July 2024 (checked 19 September 2026)) - EU, all new registrations: 7 July 2024 (Same row, date for the prohibition of registration (checked 19 September 2026)) - CSMS certificate validity: a maximum of three years (Para. 6.7, from the date of deliverance unless withdrawn; para. 6.8 lets the Approval Authority verify at any time. A cycle, not a date (checked 18 September 2026)) - ISO/SAE 21434:2021 published: August 2021 (Edition 1; the ISO catalogue shows the month only, so the month is what is carried. Under systematic review (stage 90.20) when read (checked 18 September 2026)) ## Sources - UN Regulation No 155 as published in the Official Journal (EUR-Lex, CELEX 42021X0387): https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:42021X0387 - Regulation (EU) 2019/2144, consolidated text, Annex II row D4: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02019R2144-20240707 - ISO/SAE 21434:2021 (ISO catalogue): https://www.iso.org/standard/70918.html Canonical: https://www.provael.com/compliance/un-r155-iso-21434/ --- Provael · Prove it. Prevail. · Apache-2.0 · https://github.com/provael/provael Not legal advice; verify regulatory dates against the primary source.