Cosmin Badea.
AI safety researcher, Imperial College lecturer, and the founder of Ethicos AI.
Cosmin Badea · London
Building the frameworks that let AI systems decide well — and the institutions that hold them up.
Cosmin's research builds explicit, logic-based frameworks for value-aligned decision-making, with a particular focus on interpretability for moral AI. His PhD work at Imperial College London (under Prof Marek Sergot) developed MARS, the Multi-valued Action Reasoning System, for agents that deliberate over moral rules explicitly rather than learning them implicitly.
He lectures on Contemporary Philosophy and Artificial Intelligence at Imperial, where he previously designed and led the first AI and Ethics course in the Department of Computing. He has supervised MEng and MSc research on moral AI and AI safety; several student projects have led to peer-reviewed publications.
On EU-funded Horizon programmes he leads AI and ethics on 4P-CAN, a €5.3M personalised-oncology consortium of 17 organisations across 11 countries, and serves as Chief AI Officer at FH-EARLY, a clinical early-diagnosis consortium. He is the architect of Project Nightingale, an AI-for-Good initiative focused on health and wellbeing. His published work appears in Nature Machine Intelligence, AI and Ethics, BMC Medical Ethics and elsewhere.
Outside Ethicos, he sits on advisory boards and review panels for AI ethics journals, research consortia and academic programmes. He works in English and Romanian, and reads German, French and Greek for research.
Currently held positions, and the work they involve.
Framework · PhD work, Imperial
The Multi-valued Action Reasoning System — an explicit, logic-based architecture for agents that reason about moral rules under uncertainty.
Why it matters
An AI system that can show its working.
Most modern AI infers values implicitly from training data and produces outputs no one can audit at the level of reasoning. MARS takes the opposite stance: rules and values are stated, weighed and combined in an inspectable structure, so a regulator, a clinician or a court can ask not only what the system did but why.
The framework underpins Ethicos AI's specialist advisory work on moral reasoning, decision-making under uncertainty, and interpretability for high-stakes deployment — in healthcare, in public-sector AI, and in any system whose decisions need to hold up to scrutiny.
Peer-reviewed work in moral AI, interpretability and clinical decision-making.
A small number of advisory seats each year.
Outside Ethicos AI's project engagements, Cosmin takes on a deliberately limited number of board, advisory and fractional roles each year — chosen for the seriousness of the question and the calibre of the people asking it.
Typical formats: non-executive director on technology and clinical-AI boards, fractional Chief AI / Ethics Officer for companies deploying AI at scale, investment-committee advisor for funds with serious AI exposure, and private briefings for principals, family offices and senior philanthropic programmes.
All enquiries are read personally and answered within a week. Conversations are confidential by default.
Currently open to discuss
cos@ethicos.co.uk · London