July 2026
This monthly newsletter covers recent developments and upcoming events in AI safety, ethics, and governance in Montréal.
Events
Modern Lyceum: Multi-Disciplinary Panel Discussion on AI Productivity and Workforce Transformation
Tuesday, July 7, 6:30 PM. Le Plateau-Mont-Royal, Montréal.
A multi-disciplinary panel discussion on how AI is reshaping productivity and the workforce, hosted by Samira G. Registration required.
AI Safety Speed-networking & social
Tuesday, July 14, 5:30 PM. Ω Labs, 3813 rue Saint-Denis, Montréal.
A few rounds of speed-networking to make connections quickly, followed by an open social. Whether you’re a researcher, student, policymaker, builder, or just AI-curious, come share what you’re working on and grow the local community. Hosted by Orpheus and Étienne (HΩ).
AI Safety Papers We Love #4 - Law-Following AI: Designing AI Agents to Obey Human Laws
Tuesday, August 11, 6 to 8 PM. Ω Labs, 3813 rue Saint-Denis, Montréal, and online.
Talita Dias (international lawyer; Professor of Practice, McGill; previously Partnership on AI, Chatham House, Oxford) presents “Law-Following AI” by O’Keefe, Ramakrishnan, Tay, and Winter: in high-stakes settings like government, AI agents should be designed to rigorously follow the law, loyal to their principals but only within legal bounds, with legal duties imposed directly on AI agents. Advance reading encouraged but not required.
IVADO Workshop: Governing AI: Scales, Spaces and Approaches
Monday, August 24 to Thursday, August 27, 9:30 AM to 5 PM. IVADO and HEC Montréal.
A four-day multidisciplinary workshop on the legal instruments framing AI and multi-level AI governance, part of IVADO’s thematic semester on AI governance. Registration is open.
Policy and Governance
Canada: The CUSMA review is Canada’s last chance to govern AI
In Policy Options, Tiran Rahimian Bajgiran (Law Commission of Canada, Harvard Law School) argues Ottawa is entering CUSMA review talks without AI legislation or a clear digital-trade strategy, and proposes three demands: confirming the right to adopt non-discriminatory measures governing high-risk AI systems, preventative auditing authority over algorithmic systems, and explicit language preserving provincial authority to attach data-governance conditions to data-centre energy access.
Research from Montréal
LawZero lays out a formal safety case for its “Scientist AI”
July 2. LawZero, Montréal.
The team at LawZero, the Montréal nonprofit led by Yoshua Bengio, presents a mathematical framework for a “disinterested” AI that makes honest predictions without pursuing goals of its own. The work targets “implicit agency”, goal-seeking behaviour that no one asked for, and offers a safety case that is conditional on assumptions still to be tested empirically.
Collusion Risks Among AI Reasoning Agents Justify Certification Requirements for Making Market Decisions
ICML 2026. Tara Research.
Tara Research, a nonprofit studying honesty in AI systems with a team spanning the Netherlands and Montréal, shows that reasoning agents placed in simulated pricing markets drift toward tacit collusion, and argues for certification requirements before AI agents are entrusted with market decisions.
The implicated scientist: on the role of AI researchers in the development of weapons systems
Volokhova, Hernandez-Garcia. Mila, Université de Montréal.
Presented at the AI for Peace workshop, ICLR 2026. AI researchers are “implicated subjects” in the harms caused by weapons enabled by AI technologies, through military funding and industry collaborations. The authors explore how to transfigure this implication into solidarity with victims, including through collective resistance and refusal of military contracts.
Have an event, opportunity, or article for next month? Send us a note at team@horizonomega.org.
This newsletter is by HΩ, researched and written with the assistance of AI.