The Vector Institute (French: Institut Vecteur) is a private, non-profit artificial-intelligence research institute based in Toronto, Ontario, focused primarily on machine learning and deep learning research (Source: en.wikipedia.org). It hosts researchers who hold Canada CIFAR AI Chairs, the chair programme under which Nicolas Papernot's appointment is held, and its researchers commonly hold joint appointments with Canadian universities.
Founding and funding
Vector launched in March 2017 with combined funding commitments of $135 million over its first five years, from the Government of Canada through the Canadian Institute for Advanced Research (CIFAR), the Government of Ontario through the Ministry of Economic Development, Job Creation and Trade, and nearly 40 companies across sectors of the Canadian economy (Source: vectorinstitute.ai). It was established by Brendan Frey, Geoffrey Hinton, and Raquel Urtasun, with the stated objectives of retaining and recruiting researchers in Toronto and encouraging companies to establish labs in the city; one account puts the combined public and private total received at its founding at CAD $200 million (Source: en.wikipedia.org). The institute describes itself as built on the scientific foundation established by Hinton and fellow University of Toronto faculty; Hinton has served as its chief scientific adviser since launch (Source: cifar.ca).
Vector was named a key partner in the Government of Canada's $125 million Pan-Canadian AI Strategy announced in the 2017 budget, under which CIFAR supports the institute's core activities, the Canada CIFAR Chairs in AI Science, graduate training, and the participation of chairs and trainees in national AI activities (Source: vectorinstitute.ai). It operates in partnership with the University of Toronto and other institutions.
Research recorded here
Four of the six authors of "AI Agents Enable Adaptive Computer Worms" (arXiv:2606.03811, June 2, 2026) — Jonas Guan, Tom Blanchard, Hengrui Jia, and Papernot — list joint University of Toronto and Vector Institute affiliations. The paper reports a contained proof-of-concept worm driven by an open-weight language model hosted on compromised GPUs, and argues that vendor-side safety controls cannot reach a threat that uses no vendor API (AI Agents Enable Adaptive Computer Worms (Guan et al., June 2026)). Its authors state that before publication they sought guidance through a University process and disclosed the threat of AI-driven worms to several Government of Canada entities.
Relationships
- related: Nicolas Papernot; Autonomous cyber-agents; Agentic harnesses and capability elicitation
- affiliated-with: CIFAR (Canada CIFAR AI Chairs programme); University of Toronto
- anchor-sources: AI Agents Enable Adaptive Computer Worms (Guan et al., June 2026)
Confidence
Medium. Founding, funding, and programme detail come from the institute's own site and press release corroborated by CIFAR and an encyclopedia entry; the two funding figures ($135 million in commitments over five years, CAD $200 million received at founding) are reported on different bases and are not reconciled by the sources consulted. Current headcount, budget, and research output are not documented here.