What the living guidelines are
The Living guidelines on the responsible use of generative AI in research are produced by the European Commission's Directorate-General for Research and Innovation together with the ERA Forum, which brings together Member States, associated countries and research stakeholders (European Commission, 2026). They are "living" because they are revised as the technology changes. The second version is dated April 2025 (European Commission, 2025), and the third edition is dated May 2026 (European Commission, 2026).
The guidelines are non-binding. In the Commission's words they "should be considered as a supporting tool for researchers, research organisations and research funding bodies" (European Commission, 2026). They draw their principles from the European Code of Conduct for Research Integrity and from the work on trustworthy AI of the High-Level Expert Group on AI, and they complement the AI Act rather than replacing it (European Commission, 2026).
Six recommendations for researchers
For doctoral researchers the core is a list of six recommendations: remain ultimately responsible for scientific output; use generative AI transparently; pay particular attention to privacy, confidentiality and intellectual property rights; respect applicable national, EU and international legislation; continuously learn how to use the tools properly; and refrain from using generative AI tools substantially in sensitive activities (European Commission, 2026).
Two clarifications help with daily decisions. First, authorship stays human: "AI systems are neither authors nor co-authors" (European Commission, 2026). Second, the guidelines separate substantial use, which should be disclosed, from basic editorial support or using a tool to understand a concept, which they describe as not substantial (European Commission, 2026). Your journal or university may draw that line differently, so follow the stricter rule where they differ.
What universities and funders are asked to do, and what changed in 2026
Research organisations are asked to promote, guide and support responsible use, to track how generative AI is used internally, to reference or integrate the guidelines in their research guidance, and to provide locally hosted or cloud-based tools that protect data (European Commission, 2026). Funders are asked to promote responsible use, review their own internal use, request transparency from applicants, monitor the landscape and account for the risk of hidden prompts (European Commission, 2026).
The 2026 revision is described as limited and technical. It adds recommendations on AI used by third parties, for example in meetings, note-taking, summarising discussions or managing information, and it addresses hidden prompts: instructions embedded in content that may not be visible to users or subject to human oversight (European Commission, n.d.). For a PhD student, the practical message is to be careful with AI meeting assistants in supervision or project meetings, and never to embed hidden instructions in a proposal or manuscript.
