At LawZero, we're rethinking the building blocks of frontier AI to create an intelligent machine that is both highly capable and safe-by-design. We’re excited to share our first blog post outlining some of the objectives and core components of our Scientist AI project. 🧵 (1/4)

This is your sign to leave the race pushing AI to the brink. Put your talent towards unlocking the full possibilities of AI without compromising on safety. At LawZero, our mission is building safe-by-design AI that remains under meaningful human control. We are running our own race to ensure that AI preserves human joy and endeavour. Join us. You have the power to choose a different path: job-boards.greenhouse.io/law… 📰 Read Yoshua Bengio’s full op-ed in Transformer: open.substack.com/pub/transf…

Treating recent AI incidents as simple cybersecurity problems misses the root cause: the fact that leading companies continue to drive capability forward without knowing how to guarantee alignment and safety. In his new op-ed, @Yoshua_Bengio explains how current AI training methods teach systems to pursue goals above all else. LawZero is working on a fundamental solution by developing safe-by-design and trustworthy AI systems built without autonomous preferences or misaligned objectives. It’s time we stop patching symptoms and address the architecture.

It was an honour to welcome Her Royal Highness the Infanta Cristina of Spain and Ariadna Bardolet of the “la Caixa” Foundation to the LawZero offices yesterday. We had a rich conversation on how AI can and should be developed for the benefit of humanity, and on the importance of developing safe-by-design approaches that keep AI under meaningful human control. This is exactly what we are doing at LawZero with the development of Scientist AI: taking a new approach to building highly capable AI that operates with no goals of its own. We are grateful to @FundlaCaixa for its longstanding commitment to science in the service of people, and we look forward to continuing the conversation.

Ce fut un honneur d'accueillir Son Altesse Royale l'infante Cristina d'Espagne et Ariadna Bardolet, de la Fondation « la Caixa » dans les locaux de LoiZéro hier. Nos échanges ont principalement porté sur la manière dont l’IA peut et doit être développée au service de l'humanité ainsi que sur l'importance de développer des approches techniques garantissant que les systèmes d’IA sont sécuritaires dès leur conception. C’est l’essence de notre démarche à LoiZéro alors que nous développons l’IA-chercheuse : adopter une nouvelle approche pour concevoir une forme d’IA extrêmement performante qui fonctionne sans aucun objectif propre. Nous remercions la @FundlaCaixa pour son engagement de longue date en faveur d’une science au service des citoyens, et nous avons hâte de poursuivre ce dialogue.

Guten Tag 🇩🇪 A major piece of yesterday's funding news is that we're establishing an office in Berlin, Germany! Do you want to be part of building our European team? job-boards.greenhouse.io/law…

Guten Tag 🇩🇪 Élément clé de notre annonce d'hier : nous ouvrons un bureau à Berlin, en Allemagne ! Voulez-vous contribuer à la création de notre équipe en Europe? job-boards.greenhouse.io/loi…

ALL IN 2026 — JOUR 1 ✔️ Une journée bien remplie! Entre l'annonce du partenariat Mila x @DFKI, l'annonce de financement pour @LawZero_, plusieurs conférences enrichissantes, et un pavillon Mila animé toute la journée, il y avait de quoi faire. Merci pour toutes les belles rencontres. À demain pour la suite! - @allincanada 2026 — DAY 1 ✔️ Day one brought a lot: we announced a new partnership with DFKI, celebrated a major milestone for LawZero, sat in on several thoughtful talks, and welcomed a steady stream of visitors to the Mila pavilion. Thank you to everyone who stopped by and took part in the conversations. See you tomorrow for more!

As AI gets more powerful, safety cannot be an afterthought. Today, Canada and Germany announced a combined $300 million investment in @LawZero_ , founded by @Yoshua_Bengio. Canada helped lead the world in AI research. Now, we’re helping lead the work to make AI safe, reliable and worthy of trust. 🇨🇦 À mesure que l’IA devient plus puissante, la sécurité ne peut pas être une réflexion après coup. Aujourd’hui, le Canada et l’Allemagne ont annoncé un investissement combiné de 300 millions de dollars dans LoiZéro, fondée par Yoshua Bengio. Le Canada a contribué à faire progresser la recherche mondiale en IA. Maintenant, nous contribuons à mener les efforts pour rendre l’IA sûre, fiable et digne de confiance.

Yoshua Bengio’s non-profit to get up to $300-million from Canada, Germany to expand safe AI development theglobeandmail.com/business…

AI systems become more powerful, we cannot afford to treat safety as an afterthought. Today, Canada and Germany each announced a $150 million investment in LawZero, the AI safety organization founded by Yoshua Bengio, one of Canada’s great scientists. Canada helped lead the world in AI research. Now, we are helping lead the serious work needed to make AI safe, reliable and worthy of people’s trust. 🇨🇦 À mesure que les systèmes d'IA gagnent en puissance, nous ne pouvons pas nous permettre de traiter la sécurité comme une considération secondaire. Aujourd'hui, le Canada et l'Allemagne ont chacun annoncé un investissement de 150 millions de dollars dans LoiZéro, l'organisation de sécurité en IA fondée par Yoshua Bengio, l'un des grands scientifiques du Canada. Le Canada a contribué à mener le monde en recherche en IA. Aujourd'hui, nous contribuons à mener le travail sérieux nécessaire pour que l'IA soit sûre, fiable et digne de la confiance des gens.

We are happy to share that Canada and Germany have announced funding of up to $300 million for LawZero! This will power the next phase of our technical roadmap towards developing a fundamentally new form of advanced, safe, and capable AI. lawzero.org/en/news/lawzero-…

LoiZéro a reçu un engagement de financement conjoint allant jusqu'à 300 M$ du Canada et de l'Allemagne! Cet investissement propulsera la prochaine phase de notre recherche vers une forme fondamentalement nouvelle d’IA avancée, sécuritaire et performante. lawzero.org/fr/nouvelles/loi…

Demain après-midi à la conférence ALL IN, notre fondateur @Yoshua_Bengio sera au kiosque LoiZéro pour un entretien informel. Il y partagera ses réflexions sur la sécurité de l'IA et notre direction de recherche. Posez vos questions ici : tally.so/r/yPBjY0

Tomorrow afternoon at ALL IN, our Founder @Yoshua_Bengio will be at the LawZero booth for a fireside chat to share his thoughts on AI safety and our scientific direction! Submit your questions ahead of the discussion: tally.so/r/yPBjY0

In his latest blog post, our Founder @Yoshua_Bengio shares his analysis of recent incidents involving misbehaving AI agents: a critical scientific challenge we are tackling with our research.

Retrouvez l'équipe de LoiZéro à #ALLIN2026 à Montréal la semaine prochaine ! Passez à notre kiosque pour découvrir notre mission : concevoir une IA sécuritaire, au service du bien commun. Au plaisir de vous y voir !

Heading to #ALLIN2026 in Montreal next week? Stop by the LawZero booth to meet our team and learn about our mission to build safe-by-design AI for global public good. See you there!

We were honoured to host Her Excellency the Right Honourable Louise Arbour, Governor General of Canada, at LØ! Her Excellency met with @Yoshua_Bengio and the team to discuss AI safety and Canada’s role in shaping safe-by-design AI that serves the public good and empowers humanity

Another real-world manifestation of the kind of misaligned actions frontier systems developed by leading companies can take to achieve goals. Current frontier AI models are trained with reinforcement learning to have goals and independently work out a path from A to B using planning, strategy and, occasionally, shortcuts and unintended subgoals. This misaligned behaviour has been understood theoretically and observed empirically, and the more capable an AI is, the more this misalignment can be amplified. In this case we see the type of actions this could lead to once guardrails are bypassed: creating fake identities on the web, sending targeted emails, and attempting to get malicious code integrated into an open-source project. Even though guardrails were voluntarily removed in this case, there's no guarantee that real-world guardrails will always be sufficient. On the contrary, cybersecurity is generally imperfect: companies can try patching the problem, but a sufficiently capable AI will find an unexpected loophole. Developers should be held legally responsible for damage done by their models, and greater model safety is necessary from the start, during the training phase. That’s the type of solution I'm looking into at @LawZero_. axios.com/2026/08/04/anthrop…

Global cooperation is essential to build safe and trustworthy AI. On the ProHuman podcast, our Co-President and Executive Director, Sam Ramadori, and former New Zealand PM Helen Clark discuss how nations can work together to protect humanity: foreignpolicy.com/podcasts/p…