Camps

Ethics, IHL, and Human Rights

  • Defence in an academic context
engineer-check-and-control-automation-robot-arms-machine-in-intelligent-factory-industrial-.jpg

SEIUM

At SEIUM, technology is never neutral: it is designed, tested, and transferred within an explicit ethical, legal, and human rights framework. This policy sets out what we do, how we do it, and what we do not do when working in sensitive areas—including defence in an academic context, civil security, autonomy, and dual use—to ensure that teaching and research are non-operational, socially responsible, and compliant with International Humanitarian Law (IHL) and international human rights standards.

,

Purpose, scope, and institutional positioning

At SEIUM, we do not only teach engineering: we build it alongside industry and regulators, in laboratories that can be accredited and with real-world standards. Working here means

Ensure that all academic, research, and third-party engagement activities comply with International Humanitarian Law (IHL), human rights, applicable legislation, and internal policies; minimise the risk of misuse; and protect individuals, communities, and the public interest.

 
 

Students, faculty, technical staff, administrative staff, adjuncts, fellows, guests, contractors, and partners (companies, centres, and public authorities). Applies to content, data, software, equipment, testing, publications, and technology transfer.

 
 

We do not teach or develop operational tactics, doctrine, or TTPs (tactics, techniques, and procedures); we focus on engineering, safety, certification, ethics, and validation.

 
 

We exclude the design, optimisation, or instruction of offensive capability, lethality, weapons guidance, the exploitation of vulnerabilities outside legal frameworks, or the deliberate circumvention of regulatory safeguards.

The principles of distinction, proportionality, and precaution under International Humanitarian Law (IHL); the UN Guiding Principles on Business and Human Rights; OECD due diligence guidance; and applicable national and international regulations.

Offensive capability instruction, lethality, weapons guidance, the exploitation of vulnerabilities outside legal frameworks, or the deliberate circumvention of regulatory safeguards.

Guiding principles

First, do no harm:

Preventive assessment of human, social, and environmental risks.

Legality and legitimacy

Regulatory compliance plus explicit ethical justification (“not everything legal is legitimate”).

Human-in-the-loop

Meaningful human oversight in autonomy and decision-making systems.

Transparency and traceability

Documented decisions; auditable by internal committees and, where applicable, external bodies.

Privacy and dignity

mínimos de datos, privacy-by-design, protección de colectivos vulnerables.

Fairness and non-discrimination

Data minimisation, privacy by design, and protection of vulnerable groups.

Proportionality and minimisation

The minimum necessary for a legitimate academic/scientific purpose.

Accountability

Clear responsibilities, whistleblowing channels, and corrective measures.

Governance: Roles and Bodies

Ethics and Dual-Use Committee (EDUC): An independent body that evaluates, classifies, and authorises sensitive projects and courses. It includes experts in ethics, engineering, law, human rights, security, privacy, and student observers.

Office of Compliance and Export Controls (OCE): Classifies technologies and data (ITAR/EAR/dual-use), authorises access, and coordinates licensing.

DPO/Privacy: Ensures DPIAs/PIAs where personal data or monitoring is involved, and oversees international data transfers.

HSE / Laboratory Safety: Ensures health, safety, and the physical environment of testing activities.

Academic Directorate: Does not approve teaching or research plans in sensitive areas without a formal opinion from the EDUC/OCE/DPO.

Decision Chain (Simplified)

Proposer → Supervisor/PI → EDUC (ethics/dual-use) + OCE (export controls) + DPO (privacy) → Vice-Rectorate (approval) → Execution with continuous audit.

Activity Classification

  • Green (low risk): General content, civil state-of-practice, synthetic or anonymised datasets, with no clear dual-use potential.
  • Amber (medium risk): Autonomy, perception, cyber-physical systems, abstract tactical simulation, or potential dual-use with mitigations. Requires EDUC review and access controls.
  • Red (high risk / prohibited): Lethality, weapons guidance, circumvention of safeguards, de-anonymisation, live operations, or offensive exploitation without legal authorisation. Not permitted at SEIUM.

Ethical Review Workflow (Lifecycle)

Phase 0 — Idea

Initial screening (sensitivity, human rights, and export controls checklist). Draft a Human Rights Impact Assessment (HRIA) where it may affect individuals or communities.

Phase 1 — Proposal

• Initial screening (sensitivity, human rights, and export controls checklist). • Draft a Human Rights Impact Assessment (HRIA) where it may affect individuals or communities.

Phase 2 — Design

• Mitigation plan: anonymisation/pseudonymisation, clean rooms, need-to-know segregation, export controls, rate limiting, watermarking, model cards. • DPIA/PIA if personal data is involved; de-biasing and validation plan.

Phase 3: Execution

• Progress gates with EDUC/HSE/OCE reviews. • Logging and audit trails; kill switches to stop testing.

Phase 4: Results and Dissemination

• Publication review: what is published, how, and with which safeguards. • Licensing, use clauses, and liability notices.

Phase 5: Closure and Lessons Learned

Final report, ethical post-mortem, data retention/deletion plan, lessons learned.

Design/optimisation of lethal effects or weapons guidance.

Exploitation of vulnerabilities (cyber/OT) outside legally authorised programmes and without responsible coordination.

Re-identification of individuals, or use of sensitive datasets without a legal basis/consent and enhanced safeguards.

Deliberate circumvention of regulatory safeguards (safety/certification).

Live training or realistic simulation of operational tactics intended for use in conflict.

Red Lines (Non-negotiable)

Dual-Use and Autonomy Management

  • Capability reduction: publish principles and aggregated results, not critical parameters that could enable offensive uses.

     
     

    • Synthetic/anonymised data by default: If real-world data is used, require DPIA/HRIA, data minimisation, and data contracts.

     
     

  • Responsible autonomy: human-in-the-loop/on-the-loop, operating limits, fail-safes, safety cases, and assurance cases.

     
     

  • Defensive cybersecurity: focus on hardening, detection, response, and resilience; no offensive development.

     
     

Engagement with Partners (Contracts and MoUs)

  • IHL/Human Rights clause: academic and civil use only; prohibition of offensive purposes.
  • Dual-use & export controls: prior classification; access controls by nationality and role; required licences where applicable.
  • Data and privacy: ownership, permissions, minimization, transfers, retention, non-re-identification.
  • Data and privacy: ownership, permissions, minimisation, transfers, retention, and no re-identification.
  • Audit and termination: right to audit and to terminate the agreement in case of non-compliance.

Mandatory Training

Principles, borderline cases, warning signs, red flags.

 
 
 

DPIA, anonymisation, bias, explainability, and ARCO/DSAR rights.

 
 
 

Classification, counterparty screening, licensing, workflows.

 
 
 

Safe procedures, substances/equipment, incident response.

 

Best practice, secure-by-design, incident response.

 
 
 

  • Important: 100% of staff and students participating in sensitive projects must keep their training up to date; non-compliance results in suspension of access to resources and repositories.

 
 
 

Publication, Communication, and Open Science

Preference for openness where it does not compromise people, safety, or compliance.

 
 

Synthetic data, aggregated results, omission of critical parameters.

Purpose, limitations, intended use, out-of-scope uses, residual risks.

 
 

Avoid overclaiming; contextualise risks and safeguards.

 
 

SEIUM

Assume that technical excellence without responsibility is not excellence. For that reason, all teaching and research are governed by a robust framework for ethics, International Humanitarian Law (IHL), and human rights, with clear red lines, independent committees, auditable processes, and a “stop the line” culture. Our commitment is to educate and transfer engineering that is safe, lawful, and useful to society—from the road to orbit—without crossing the limits that protect life, dignity, and the rule of law.

Incidents, Reports, and Sanctions

  • Confidential reporting (whistleblowing) channel with whistleblower protection and an SLA of ≤72 hours.
  • Response protocol: containment, investigation, notification (where applicable), remediation, and lessons learned.
  • Sanctions regime: from access withdrawal and academic suspensions to contract termination or reporting to authorities, depending on severity.

Metrics and Transparency

  • % of projects classified (green/amber/red).
  • Number of EDUC reviews and resolution times.
  • % of projects with an HRIA/DPIA and effective mitigations.
  • Compliance training up to date (target: 100%).
  • Critical incidents (goal: 0).
  • Publications with model/data cards and dual-use disclosures.
  • Annual public report with aggregated indicators and improvement actions.

Illustrative Cases (Practical Guide)

Operational Annexes (Available on the Internal Portal)

Permitted with safeguards (amber):

1. Obstacle-avoidance algorithms for civil UGVs: synthetic data, human-on-the-loop, a safety case, and limits on speed and operating environment.
2. GNSS/INS resilience evaluation for logistics: testing in a controlled testbed, without degrading public safety.

Permitted (green):

1. Cabin design focused on ergonomics and HMI for prolonged operation; UX and human factors.
2. Academic-level RAMS and certification methodologies (UNECE/EASA/IMO/IAEA).

Not permitted (red)

1. Terminal guidance optimisation for munitions: no-go. 2. Exploitation of industrial PLCs without a legally authorised, coordinated research framework: no-go.
  • Ethical and Dual-Use Sensitivity Checklist (v1.3)
  • Ethics Dossier Template (HRIA + DPIA Lite)
  • Green/Amber/Red Classification Matrix
  • Sensitive Data and De-identification Guide
  • Standard Contract Clauses (IHL/Human Rights, Export, Data, Publication)
  • Responsible Publication Guide (Model/Data Cards)
  • Incident Procedure and After-Action Review
Scroll to Top