Research institute · Research orgs · Montréal, Québec · Est. 2025
LawZero is a Montréal-based nonprofit research organization focused on “safe-by-design” advanced AI. Its core technical idea is “Scientist AI”: a system intended to support honest, evidence-based prediction and reasoning while reducing the likelihood of goal-directed behavior and related safety failures seen in more agentic systems.
LawZero frames Scientist AI as a multi-component approach aimed at disentangling understanding from preferences over how the world unfolds, and as a potential “guardrail” for frontier AI agents. In addition to developing the technical approach, LawZero positions its work as contributing a global public-good pathway for advancing AI safely and responsibly, insulated from market pressures that it argues can prioritize deployment speed over trustworthiness.
No curated programs yet. Check the careers page ↗.
<span>Safety from Honesty in a Disinterested AI Predictor</span> <span><span>VPcomm</span></span> <span><time datetime="2026-07-02T07:00:00-04:00" title="Thursday, July 2, 2026 - 07:00">Thu, 07/02/2026 - 07:00</time> </span>
Error 403<span>Error 403</span> <span><span>mlanteigne</span></span> <span><time datetime="2025-05-13T08:13:51-04:00" title="Tuesday, May 13, 2025 - 08:13">Tue, 05/13/2025 - 08:13</time> </span>
Yoshua Bengio expressed pride in what the LawZero team is building and gratitude for landmark support from a government, though the supplied post is truncated and does not identify the government or the nature of the support.
LawZero receives a commitment of up to $300M in joint funding from Canada and GermanyCanada and Germany announced a combined commitment of up to CAD $300 million for LawZero to develop advanced, safe-by-design AI, including expansion of its research team and computing infrastructure and operations into Europe (Berlin office).
LawZero advances safe-by-design AI with support from NVIDIALawZero announced it received a donation of computing resources from NVIDIA to support research and development of Scientist AI.
An AI that Predicts but has no Hidden Agenda: LawZero Lays out a Formal Safety Case for its “Scientist AI”.LawZero released a formal safety case describing a mathematical framework for a “disinterested” AI predictor intended to produce honest predictions without pursuing goals.
LawZero to Attend ICML 2026 in Seoul, Expanding its Global Safe AI Research PresenceLawZero announced it would exhibit at ICML 2026 and that members of its research team would engage with the global AI community and recruit for safe-by-design research roles.
Rt. Hon. Dame Jacinda Ardern, Former Prime Minister of New Zealand, joins LawZero’s Global Advisory CouncilLawZero announced Jacinda Ardern joining its Global Advisory Council.