Human responsibility
LLMs should act as collaborators within certifiable workflows, not as autonomous developers without any human control.
Manifesto · CPS & GenAI · v1.0
Human-First CPS & GenAI Engineering
The Human-First Manifesto for Safe and Responsible CPS & GenAI Engineering.
A shared commitment to how we design, ship, and steward cyber-physical and GenAI-enabled systems that impact real people.
PILLARS
03
REVISION
v1.0
STATUS
LIVE
This is a living document. Edit it, fork it, bring it into the rooms where decisions happen.
Core Principles
The three anchors for every decision we make.
Independent eyes
Assurance must become an interactive, human-in-the-loop activity rather than a static, post hoc exercise.
Human-first
Ethical, legal, and societal concerns should be integrated into CPS development and operations from the outset, not treated as downstream constraints.
Manifesto
A declaration of intent for the responsible integration of GenAI in the engineering of safety-critical CPSs.
The emergence of GenAI presents new opportunities and challenges in the engineering of safety-critical CPSs. These systems operate in high-assurance environments where reliability, certification, and accountability are essential. GenAI can support development processes, but its use must maintain human authority, preserve certification pathways, and ensure trustworthiness across the entire system lifecycle.
We affirm three core principles: human responsibility, certification integrity, and transparent integration. GenAI should augment, not replace engineers in roles where safety, compliance, and verification depend on traceable human oversight. Full lifecycle transparency, including documentation of GenAI involvement, is necessary to sustain regulatory alignment and public trust.
We commit to the safe, auditable, and standards-aligned use of GenAI in critical CPSs. Our shared goal is not only to advance engineering capability, but to ensure that these systems continue to meet the highest expectations of safety, accountability, and societal value.
Original scientific publication
This manifesto is grounded in the following peer-reviewed publication.
Research & Practice Agenda
A working list of thematic areas where research and practice must evolve to realize this manifesto.
Human-Controlled Design with GenAI
Ensuring that every AI-assisted design activity strengthens—not replaces—human responsibility, traceability, and control throughout the CPS lifecycle. This includes collaborative design environments, socially traceable artifacts, and tools that give engineers explicit control over AI behavior.
Human-Centric Assurance and Oversight
Reimagining verification, validation, and certification as continuous, interactive processes where humans maintain independent oversight. This involves dynamic assurance, adaptive certification methods, and transparent pipelines that evolve alongside AI-generated artifacts.
Ethical, Legal, and Societal Values by Design
Embedding human-first principles directly into development workflows. This requires value-sensitive engineering, governance models involving multiple stakeholders, and ethical assurance cases that align system behavior with societal trust and normative responsibility.
Timeline
-
December 2025
Manifesto Created
The Human-First Manifesto for Safe & Responsible CPS & GenAI Engineering is drafted and released to the community.
The Community
A global coalition committed to shaping GenAI and critical systems deliberately, transparently, and always with humanity at the center.
Who this is for
- People who design, certify, operate, and govern safety-critical systems.
- Researchers, engineers, policymakers, and industry leaders shaping the role of GenAI in society.
- Anyone who believes that when human lives are at stake, “good enough” is never enough.
How we operate
- We collaborate openly: across disciplines, organizations, and perspectives.
- We challenge assumptions: subjecting GenAI-enabled systems to rigorous, independent evaluation.
- We commit to purpose: ensuring every system we build serves people, responsibility, and societal well-being.
Current supporters
6 people and 4 organisations have committed to this manifesto.
| Name | Affiliation | Country |
|---|---|---|
| Alessio Bucaioni | Mälardalen University | Sweden |
| Antonio Cicchetti | Mälardalen University | Sweden |
| Gordana Dodig Crnkovic | Mälardalen University | Sweden |
| Romina Spalazzese | Mälmo University | Sweden |
| Emma Söderberg | Lund University | Sweden |
| Dániel Varró | Linköping University | Sweden |
This is a partial list. The list will grow as the community expands.
