

ALICE x AI Circle | NeurIPS Paris Private Dinner & Expert Panel: Safety by Design
Expert panel and Private closing dinner for NeurIPS Week, hosted by Alice and AI Circle
Building safe AI is one thing. Proving it's safe when the real world gets involved? That's an entirely different challenge.
As AI systems become more autonomous, the conversation around safety is shifting. It's no longer enough to build capable models and add safeguards before deployment. Researchers and engineers are being asked to think about safety from the architecture up, while navigating real-world constraints around performance, latency, security, and increasingly complex regulatory expectations.
And in Europe, the stakes are particularly interesting.
With the EU AI Act reshaping expectations around AI development, transparency, and accountability, a bigger question emerges: Are these new constraints pushing us toward genuinely safer, more reliable AI systems, or simply better-documented ones?
During NeurIPS week in Paris, Alice and AI Circle are bringing together researchers, engineers, founders, and technical leaders for an evening exploring what it actually takes to make AI safety work beyond the benchmark.
From adversarial testing and runtime guardrails to agentic systems and regulatory accountability, we'll move past the theoretical and into the engineering decisions shaping the next generation of AI.
What we'll explore
Safety by Design, Not by Default: What does building safety into an AI system actually look like before the first model is deployed?
The European Experiment: Is the EU AI Act driving better engineering practices, or creating a new layer of compliance complexity?
The Cost of Guardrails: How do teams balance safety, performance, and latency without compromising the user experience?
When Agents Take Action: What changes when AI systems can access tools, execute tasks, and make decisions beyond generating text?
The Evening
6:00–6:30 PM | Arrival, drinks & introductions
6:30–7:15 PM | Expert panel discussion
7:15–9:00 PM | Happy hour & networking
Who Should Join?
This evening is designed for AI researchers, ML and platform engineers, safety and security specialists, founders, and technical leaders working at the intersection of AI development, deployment, and trust.
Whether you're designing agent architectures, evaluating model behavior, building adversarial testing frameworks, or deciding what needs to happen before your next production launch, this is a conversation worth being in the room for.
Following the discussion, we'll trade slides for drinks and continue the conversation with the people navigating these challenges firsthand.
Thoughtful debate. Strong technical perspectives. Great company.
And hopefully, a few disagreements worth having.
Friday, December 11, 2026
Paris, France
Space is limited. Registration is subject to approval.
About Alice
Alice (formerly ActiveFence) is an AI security, safety, and trust company helping organizations test, protect, and monitor AI systems throughout their lifecycle, from development through real-world deployment.
About AI Circle
AI Circle is a global community connecting researchers, engineers, founders, and technical leaders advancing artificial intelligence through curated gatherings, technical programming, and meaningful collaboration.