–
Coffee and Check-In
–
Welcome Address
–
Simulating Counterfactual Training
Roger Grosse (University of Toronto)
–
Break
–
AI Safety Via Inference-Time Compute
Boaz Barak (Harvard University)
–
Lunch (on your own)
–
Controlling Untrusted AIs With Monitors
Ethan Perez (Anthropic)
–
Break
–
Game Theoretic Approaches to AI Safety
Georgios Piliouras (Google DeepMind + SUTD)
–
Reception
–
Coffee and Check-In
–
Full-Stack Alignment
Ryan Lowe (Meaning Alignment Institute)
–
Break
–
Can We Get Asymptotic Safety Guarantees Based On Scalable Oversight?
Geoffrey Irving (UK AI Safety Institute)
–
Lunch (on your own)
–
Amortised Inference Meets Llms: Algorithms And Implications For Faithful Knowledge Extraction
Nikolay Malkin (University of Edinburgh)
–
Break
–
Superintelligent Agents Pose Catastrophic Risks — Can Scientist AI Offer a Safer Path? | Richard M. Karp Distinguished Lecture
Yoshua Bengio (IVADO - Mila - Université de Montréal)
–
Panel Discussion
Yoshua Bengio (IVADO - Mila - Université de Montréal), Dawn Song (UC Berkeley), Roger Grosse, Geoffrey Irving, Siva Reddy (IVADO - Mila - McGill University)
–
Coffee and Check-In
–
Robustness of jailbreaking across aligned LLMs, reasoning models and agents
Siva Reddy (IVADO - Mila - McGill University)
–
Break
–
Adversarial Robustness of LLMs' Safety Alignment
Gauthier Gidel (IVADO - Mila - Université de Montréal)
–
Break
–
Antidistillation Sampling
Zico Kolter (Carnegie Mellon University)
–
Lunch (on your own)
–
Causal Representation Learning: A Natural Fit for Mechanistic Interpretability
Dhanya Sridhar (IVADO + Université de Montréal + Mila)
–
Break
–
Out Of Distribution, Out Of Control? Understanding Safety Challenges In AI
Aditi Raghunathan (Carnegie Mellon University)
–
Coffee and Check-In
–
LLM Negotiations And Social Dilemmas
Aaron Courville (IVADO + Université de Montréal + Mila)
–
Break
–
Scalably Understanding AI With AI
Jacob Steinhardt (UC Berkeley)
–
Lunch (on your own)
–
Future Directions In AI Safety Research
Dawn Song (UC Berkeley)
–
Break
–
What Can Theory Of Cryptography Tell Us About AI Safety
Shafi Goldwasser (Simons Institute, UC Berkeley)
–
Assessing The Risk Of Advanced Reinforcement Learning Agents Causing Human Extinction
Michael Cohen (UC Berkeley)
–
Coffee and Check-In
–
Safeguarded AI Workflows
David Dalrymple (MIT)
–
Break
–
AI Safety: LLMs, Facts, Lies, And Agents In The Real World
Christopher Pal (IVADO + Polytechnique Montréal + Université de Montréal + Mila)
–
Break
–
Measurements For Capabilities And Hazards
Dan Hendrycks (Center for AI Safety)
–
Lunch (on your own)
–
Theoretical And Empirical Aspects Of Singular Learning Theory For AI Alignment
Daniel Murfet (Timaeus)
–
Break
–
Probabilistic Safety Guarantees Using Model Internals
Jacob Hilton (Alignment Research Center)
–
Closing Remarks