Mila — Quebec AI Institute

What it does

Research institute founded by Yoshua Bengio in 1993, uniting researchers from Université de Montréal, McGill, Polytechnique Montréal and HEC Montréal — 1,400+ researchers and 140+ affiliated professors. AI safety and governance work runs through an AI Governance, Policy and Inclusion division, the AI Policy Compass programme for policymakers, TRAIL responsible-AI training, and the AI4Good Lab, plus Indigenous-AI work.

Honest assessment

Several real published inboxes, which is good. But it is a large general ML institute, not a safety organisation — most of its 1,400 researchers do capabilities work. Route safety concerns to LawZero or to Bengio's academic page rather than to Mila's general inbox.

Concerns it covers

Broad. Bengio's presence anchors serious loss-of-control work, while the institutional programmes emphasise responsible AI, inclusion and policy capacity. Do not assume Mila as an institution shares Bengio's x-risk position — most of it does not.

Government channel

Strong in Canada — Mila researchers advise the Canadian government and Bengio chairs both the International AI Safety Report and co-chairs the UN Scientific Panel.