The Misalignment Sandbox
About the event
Event Overview With AI systems becoming more capable and autonomous, the prospect of advanced AI causing human extinction makes understanding their behaviour one of the most consequential research challenges of our time. Can a virtual world help us test their moral boundaries before real-world failures do? Following the May 2026 announcement of Google DeepMind taking a minority equity stake in Fenris Creations and entering into a research partnership, the two teams have been exploring the EVE Universe as a testing ground for advanced AI. The work spans EVE Online and EVE Frontier, investigating how autonomous agents choose to behave in a persistent world of shifting alliances, competing interests and complex moral choices. Will an AI honour an agreement when resources run scarce? Resist manipulation? Recognise when an action is possible, but wrong? Join us in a private penthouse suite during Singapore’s TOKEN2049 week for an invitation-only conversation with the people driving this research, alongside a small group of investors, policymakers, researchers and industry leaders. Explore the research with Google DeepMind Speakers Hilmar Veigar Pétursson — CEO, Fenris Creations Leading the studio since 2004, Hilmar has helped pioneer the player-driven economies and persistent societies that make EVE a compelling environment for AI research. Alexandre Moufarek — Director, Inception, Google DeepMind Alexandre works on breakthrough AI experiences, with contributions spanning Project Astra, SIMA, Genie and Gemini, and a background in games and robotics. Áslaug Arna Sigurbjörnsdóttir — Senior Director, AI Partnerships, Fenris Creations Iceland’s former Minister of Justice and former Minister of Higher Education, Science and Innovation now leads Fenris’ relationships with AI labs, universities and policymakers. Venue
The Misalignment Sandbox is hosted at Singapore's most premium penthouse suite, in an iconic 5-star hotel overlooking Marina Bay. With floor-to-ceiling views…
