top of page

Past Events

Here are some of the events we've hosted recently!

Launch Party 1.jpg

27 Feb 2025

Launch Party

We kicked things off with a vibrant launch party attended by over 70 participants! Our members delivered lightning talks, sharing insights into the exciting work they're doing.

13 March 2025

International AI Governance: Key Challenges and Solutions

We welcomed Su Cizem, a visiting analyst from the Centre pour la Sécurité de l'IA (CeSIA), who gave a thought-provoking talk on "International AI Governance: Key Challenges and Solutions."

Su Cizem - International AI governance.jpg
Elliot Thornley's talk.jpg

20 March 2025

Can incomplete preferences keep artificial agents shutdownable?

Elliott Thornley, visiting research fellow from Oxford University, delivered a compelling talk on "Can Incomplete Preferences Keep Artificial Agents Shutdownable?" exploring challenges in AI safety and control.

27 March 2025

Paper Club: AI Control — Safety Beyond Alignment

This session focused on the topic of control evaluations - methods that test whether safety protocols remain robust even when AI systems actively try to circumvent them.

Paper club - AI control.jpg
AI Control Hackathon.jpg

28 March 2025

AI Control Hackathon 2025

We were proud to serve as the Singapore jam site for the AI Control Hackathon 2025, hosted by APART Research. This global hackathon focused on building techniques to ensure AI systems remain secure—even when they try to subvert safeguards.

17 April 2025

Paper Club: Looking inside Claude’s “Brain”

We explored Anthropic’s groundbreaking paper, "On the Biology of a Large Language Model," which uses neuroscience-inspired tools to investigate how Claude 3.5 Haiku "thinks."

Paper club - claude's brain.jpg
ICLR Networking social.JPG

27 April 2025

AI Safety Networking Social

Together with Concordia AI, FAR.AI, and Safe AI Forum, we co-hosted a networking social that brought together over 130 global attendees (including researchers, builders, and policymakers) committed to advancing safe AI.

bottom of page