| 09:00~09:45 | Registration |
| 09:45~09:55 |
Welcome Remarks
 Myuhngjoo Kim Head of Korea AISI |
| [Session 1] Foundations of AI Safety & Security Standards | |
| 09:55~10:10 |
 The Role of Standards in Safety and Security in AI Peter Mattson MLCommons & Google |
| 10:10~10:25 |
 Introduction of MLCommons Multi-regional Safety Benchmark Lora Aroyo Google DeepMind & MLCommons AIRR Multimodal WS Lead |
| 10:25~10:40 |
[Lightning talks] Reginal Partners Talks in MLCommons Multi-regional Safety Benchmark
 [Korea Team] Boryoung Kim Seoul National University  [Singapore Team] Sita Rajagopal Infocomm Media Development Authority  [Taiwan Team] Hua-Rong Chu Chunghwa Telecom Laboratories |
| 10:40~11:00 | Break Time |
| 11:00~11:30 |
[Panel Discussion] Multilingual & Cultural approaches to benchmarking
[Moderator]
Lora Aroyo Google DeepMind & MLCommons AIRR Multimodal WS Lead  Jia Yi Goh GovTech Singapore  Eugenia Kim Microsoft  Soojung Ryu Seoul National University |
| [Session 2] Governance | |
| 11:30~11:45 |
 Who Defines “Good” AI? Rethinking Evaluation Beyond Static Benchmarks Minsuk Kahng Yonsei University |
| 11:45~12:15 |
[Panel Discussion]Sovereign AI and National Strategies: Challenges and Opportunities
[Moderator]
 Jehyun Lee KIST  Youchul Kim LG AI Research  Mark Pereira AI Singapore  Seyeob Kim Datumo Inc |
| 12:15~12:45 |
[Panel Discussion]Global Trends in AI Governance
[Moderator]
Peter Mattson MLCommons & Google  Sangbeom Ham Microsoft Corporate Standards Group  Alice Oh KAIST & Presidential Council on National Artificial Intelligence Strategy  Sabina Nong Future of Life Institute  Satoshi Sekine NII, Japan |
| 12:45~13:45 | Lunch Break |
| [Session 3] The Next - Emerging Safety Challenges | |
| 13:45~14:00 |
 Frontier AI Safety and Security: Agent, Multimodal, and Physical AI Haon Park AIM Intelligence |
| 14:00~14:15 |
 Evaluating Data-Leakage Risks in Tool-Using LLM Agents: A Singapore–Korea AISI Bilateral Testing Hankyul Baek Korea AISI |
| 14:15~14:30 |
 Grounding AI Evaluation Locally with Multicultural and Multilingual Benchmarks Jun Yun Tee AI Singapore |
| 14:30~14:45 |
 Building Safe AI for a Global Userbase Shayne Longpre Anthropic |
| 14:45~15:00 |
Closing Remarks
Peter Mattson MLCommons & Google Lora Aroyo Google DeepMind & MLCommons AIRR Multimodal WS Lead |
| 15:00~16:00 |
[Closed-Door Roundtable] Online Safety Dialogue (Invitation Only)
|
| 09:00~10:00 | Registration and Coffee-Network |
| FLI AI Safety Index - Summer 2026 [with FLI] | |
| 10:00~10:10 |
Welcome Remarks
 Minnseok Choi Director of Policy/External Cooperation, Korea AISI |
| 10:10~10:20 |
Opening Remarks
 Scott Choi Korea Representative of Future of Life Institute |
| 10:20~10:30 |
The AI Safety Index: From Accountability to a Safer Frontier Sabina Nong Future of Life Institute |
| 10:30~10:40 |
 Challenges and Approaches in Regional-Language AI Safety Evaluation: A Korean Case Study Sungpil Shin Korea AISI & Chair of PG1005 |
| 10:40~10:50 |
 NAVER's Service-Centric AI Safety Framework Daeseop Song NAVER |
| 10:50~11:50 |
[Panel Discussion] Racing to the Top: Raising the Bar on Frontier AI Safety Management
[Moderator]
Sabina Nong Future of Life Institute  Sharon Li ICML 2026 Organizing Committee & University of Wisconsin - Madison  Adam Gleave FAR.AI Sungpil Shin Korea AISI & Chair of PG1005  Elad Hazan Google AI Princeton |
| 11:50~12:00 |
Closing Remarks
Alice Oh KAIST & Presidential Council on National Artificial Intelligence Strategy
|
| 12:00~13:30 | Lunch Break |
| Red-Teaming Workshop [with AIM Intelligence] | |
| 13:30~13:40 |
Session Introduction
Haon Park CTO of AIM Intelligence |
| 13:40~13:55 |
 Judgement Day: Insights from Multimodal Red-Teaming Challenge Dasol Choi AIM Intelligence |
| 13:55~14:20 |
 Shift Happens: Robustness and Reliability of Multimodal Foundation Models Jenny Ni Google  Naman Goyal Google Deepmind |
| 14:20~14:35 |
Adaptive Red Teaming for Generative AI Safety and Security Eugenia Kim Microsoft |
| 14:35~14:50 |
 Ready to Evaluate: Building Korea’s Frontier AI Evaluation Framework Sang Seo Korea AISI |
| 14:50~15:05 | Break Time |
| 15:05~15:25 |
 Automated Discovery of State-of-the-Art Adversarial Attack Methods for LLMs Maksym Andriushchenko ELLIS Tübingen & MPI-IS |
| 15:25~15:55 |
[Panel Discussion] The Next Frontier of Vulnerabilities: From Universal Jailbreaks to Multilingual Agent
[Moderator]
Haon Park AIM Intelligence  Myoungshin Kim LG AI Research  Wan Sie Lee Singapore AISI Maksym Andriushchenko ELLIS Tübingen & MPI-IS |
| 15:55~16:15 |
Red-Teaming at the Frontier: Finding Universal Jailbreaks in State-of-the-Art Models Adam Gleave FAR.AI |
| 16:15~16:30 | Break Time |
| 16:30~18:00 |
Live Red-Teaming Challenge, "Judgement Day: Jailbreaking the Humanoid" [with AIM Intelligence]
* Please refer to the "Notice" board for competition operations and instructions.
|