AI Safety Thursdays - Formal Containment: Proof-Carrying Code and AI Safety
- Date
- 2025-06-26
- Location
- Enter the main lobby of the building and let the security staff know you are here for the AI meetup. You may need to show your RSVP on your phone. You will be directed to the 12th floor where the meetup is held. If you have trouble getting in, give Juliana a call at 647-544-0993., Toronto, ON, Canada
- Host
- Trajectory Labs
About this event
If you care about building advanced AI systems that stay inside clear, enforceable boundaries, this meetup is for you. AI Safety Thursdays - Formal Containment: Proof-Carrying Code and AI Safety brings a technical safety topic into a community setting where you can learn, ask sharp questions, and meet other people thinking seriously about alignment, control, and robust system design. About the Event This is an in-person Toronto meetup focused on formal containment and proof-carrying code as they relate to AI safety. The theme points toward a practical and deeply relevant question: how do we create systems whose behavior can be constrained, verified, or checked in ways that go beyond trust, intuition, or post-hoc monitoring? Expect a discussion space that sits at the intersection of AI safety, autonomy, and technical rigor. Whether you come from machine learning, software engineering, formal methods, security, philosophy, policy, or independent research, the event is designed to create a shared frame for thinking about how strong guarantees might matter in the age of increasingly capable models. AI Safety Thursdays suggests an ongoing community rhythm, so this event is not just a one-off talk. It is also a chance to plug into a local group of people who are actively tracking hard questions in AI and want to work through them together in person. What to Expect You can expect a meetup format that combines learning, discussion, and networking. The core topic, formal containment, naturally invites both conceptual explanation and debate about implementation: what counts as a meaningful guarantee, where proof-based approaches help, and where they may be limited when applied to real-world AI systems. Likely activities at an event like this include: A focused presentation or guided discussion on proof-carrying code and its relevance to AI safety Time to unpack the idea of containment in practical, technical terms Questions from attendees with different levels of background Informal conversation before, during, or after the main session Opportunities to meet others interested in AI, autonomy, governance, and technical safety work Because the topic is specific, the conversation should be more substantive than a general-interest AI meetup. You should come ready to think carefully about verification, constraints, system boundaries, and what it would take to make safety properties legible rather than assumed. At the same time, this is still a community event, not a closed-door seminar. If you are curious and willing to engage seriously, there is room to listen, learn, and participate even if proof-carrying code is new to you. Why Attend This meetup offers a strong reason to show up if you are tired of vague AI safety conversations. The subject matter pushes toward concrete mechanisms: not just whether systems should be safe, but how one might design architectures, interfaces, or execution environments that support stronger guarantees. You may leave with a clearer understanding of: What proof-carrying code means in plain terms Why formal containment is relevant to advanced AI systems How verification-oriented thinking differs from softer safety approaches Where formal methods may fit into broader alignment and control strategies What open questions still make this area difficult and worth discussing There is also real value in the people in the room. Events like this are often where researchers, builders, and serious learners find collaborators, reading partners, future meetup regulars, or simply others who share the same level of concern and curiosity about technical AI safety. If you are looking to sharpen your thinking, pressure-test your assumptions, or get more involved in the Toronto AI safety community, this is a strong entry point. Practical Details The meetup is in person in Toronto, Canada on Thursday, June 26 at 6:00 PM EDT. This is a good fit if you want face-to-face discussion rather than another online session. To get in, enter the main lobby of the building and let the security staff know you are there for the AI meetup. You may need to show your RSVP on your phone. From there, you will be directed to the 12th floor, where the meetup is held. If you have trouble accessing the building, you can call Juliana at 647-544-0993. Since entry involves lobby check-in and security direction, it is smart to arrive a little early so you can get upstairs before the event gets underway. Bring your curiosity, your questions, and, if you like, your strongest disagreement about what meaningful AI containment should look like. This topic benefits from people who are thoughtful, technically engaged, and willing to examine both the promise and the limits of formal approaches.
Who should attend
This will be a strong fit if you want a more serious, technically grounded AI safety conversation than you usually get at broad AI events. - You work in **machine learning, software engineering, security, or formal methods** and want to explore how verification-style thinking could apply to advanced AI systems. - You are an **AI safety researcher, student, or independent learner** looking to connect alignment questions with concrete mechanisms like containment and proof-based guarantees. - You are curious about **how to make system behavior more legible, enforceable, and auditable**, not just monitored after the fact. - You enjoy meetups where the discussion goes beyond headlines and into **architecture, constraints, and technical tradeoffs**. - You want to meet other people in Toronto who care about **AI safety, autonomy, and responsible system design** and are interested in building an ongoing local community. - You do not need to be an expert in proof-carrying code already, but you should be comfortable engaging with a focused topic and asking thoughtful questions.