Rokas Gipiškis represented AI Standards Lab at the Technical AI Safety Conference in Oxford, presenting three papers co-authored with AISL team member Ze Shen Chin, other colleagues, and external research collaborators. The conference was organised by Noeon Research and the Oxford Martin AI Governance Initiative, and covered model evaluations, safety engineering, and global AI governance developments.
1. Principles and Guidelines for AI Incident Monitoring and Reporting
Identifies gaps in current AI incident monitoring and reporting practices, and proposes five principles for post-deployment monitoring and five for incident reporting, supported by a template that accommodates uncertainty and allows iterative submissions.
Paper Authors: Harleen Kaur Sidhu, Rebecca Scholefield, Nour Annan, Kevin Hernandez, Isabel Nieh Hou, Abdulrahman Alshaikhi, Ze Shen Chin, Rokas Gipiškis
2. The Case for ESM3 as a General-Purpose AI Model with Systemic Risk Under the EU AI Act
Examines whether frontier biological foundation models such as ESM3 fall under the EU AI Act’s obligations for general-purpose AI models with systemic risk, and proposes remedies where current classification falls short.
Paper Authors: Taro Qureshi, Jacob Griffith, Koen Holtman, Marcel Mir Teijeiro, Ze Shen Chin, Rokas Gipiškis
3. Principles and Guidelines for Randomized Controlled Trials in AI Evaluation
Introduces a framework for standardising AI evaluation RCTs (human uplift studies), extending the Shadish et al. (2002) four-validity framework with a fifth principle on transparency, repeatability, and verification, operationalised into 33 guidelines.
Paper Authors: Christopher Kelly, Angelica Chowdhury, Alexandra Campili, Bimpe Ayoola, Devin Barbour, Thomas Chen Dawson, Ze Shen Chin, Rokas Gipiškis