General principles for the use of AI at CERN (home.web.cern.ch)

CERN has published a set of technology‑neutral general principles governing the use, development and deployment of AI across its facilities, following approval of a CERN‑wide AI strategy. The rules apply to AI embedded in devices, cloud services and software procured or brought on‑site, as well as to in‑house projects and personal tools. They cover both scientific/technical contexts—data analysis, anomaly detection, simulation, predictive maintenance and accelerator or detector optimisation—and productivity/administrative uses such as drafting, translation, coding assistants and workflow automation. The principles require documented transparency and explainability, preserve human responsibility and oversight, and mandate legality, non‑discrimination and “do no harm” practices. They also emphasize security and safety (confidentiality, integrity, availability), data privacy, sustainability, and a strict non‑military purpose. Practically, this means AI deployments at CERN must be lifecycle‑managed, auditable, risk‑assessed for bias, environmental and social impact, and subject to human validation and cybersecurity protections. For the AI/ML community, the framework signals CERN’s intent to balance scientific innovation with ethical, legal and operational safeguards—setting expectations for procurement, research workflows and collaborations that use or contribute AI capabilities.