Research integrity in the AI era: the agenda for a June 2026 Switzerland-China workshop

What research integrity means

Research integrity concerns how knowledge is produced, checked and communicated. The Swiss code of conduct for scientific integrity places reliability, honesty, respect and accountability at its centre. Developed with research, university and funding organisations, it provides a common reference for responsible scientific work.

AI makes these principles concrete in new ways. A tool might help analyse data, improve a manuscript or organise a literature search. The researcher still needs to understand and check the result. The Swiss National Science Foundation (SNSF) identifies traceability and reproducibility among the questions raised by AI, alongside protecting confidential information. Its guidance on responsible AI use keeps accountability with researchers and reviewers.

What the June agenda covered

Swissnex listed the fourth Sino-Swiss Research Integrity Workshop for 26 June 2026, jointly hosted by Swissnex in China, the Swiss embassy and Frontiers. The published programme ranged from research-integrity frameworks to research assessment, access to knowledge and data governance. It asked how AI should be used in research and publication, and how funders and institutions should evaluate researchers and their work.

This breadth matters: integrity is not confined to the final paper. It also concerns the systems that select, evaluate and distribute research. That is our interpretation of the agenda, rather than a conclusion attributed to participants.

Why the Switzerland-China connection matters

The programme sat within an existing strand of bilateral exchange. Swissnex's September 2024 Tech&Ethics programme also brought AI, governance and ethics together, listing a second research-integrity workshop and an intercultural dialogue. Its questions included trustworthy AI, safety, societal values and AI's influence on academic knowledge production.

For collaborators, these agendas offer a useful conversation starter. Working on the same project does not itself settle how partners disclose AI use, handle unpublished material or assess contributions. Discussing those expectations early is a practical lesson we draw from the themes, not evidence that the two countries have adopted identical rules.

From principles to working decisions

A concrete Swiss example appears in the SNSF's August 2026 update for peer reviewers. Reviewers must first form an independent assessment. AI can assist their work, but cannot replace their expertise. Confidential proposal material must remain protected, and AI use must be disclosed. These are SNSF review conditions, not a bilateral rule or a result of the June workshop.

For a Switzerland-China research team, we suggest three questions: which AI-assisted tasks require checking, what information may enter a tool, and how each partner will record its use. The answers should be checked against the relevant institution, funder and publisher requirements. A workshop agenda can identify common questions; it cannot supply every project's operating instructions.