Tessera

𝗡𝗢𝗨𝗩𝗘𝗔𝗨 𝗕𝗟𝗢𝗚𝗨𝗘 𝗣𝗹𝗮𝗶𝗱𝗼𝘆𝗲𝗿 𝗽𝗼𝘂𝗿 𝗹'𝗜𝗔-𝗖𝗵𝗲𝗿𝗰𝗵𝗲𝘂𝗿 👇 #LoiZéro #IAResponsable #LLM

LoiZéro | Plaidoyer pour l'IA-Chercheur

Dans nos blogues précédents, nous avons exposé deux problèmes qui affectent les grands modèles de langage actuels. Le premier est la flagornerie, soit la tendance d'un modèle à dire aux utilisateurs c...

lawzero.org

𝗡𝗘𝗪 𝗕𝗟𝗢𝗚 The Case for Scientist AI 👇 #LawZero #LLM #SafeAI #ScientistAI

LawZero | The Case for Scientist AI

In previous posts, we outlined two problems besetting contemporary LLMs. The first is sycophancy, a model's tendency to tell users what they want to hear at the expense of factual accuracy. The second...

lawzero.org

𝗡𝗘𝗪 𝗕𝗟𝗢𝗚 𝗧𝗿𝗮𝗶𝗻𝗲𝗱 𝘁𝗼 𝗽𝗹𝗲𝗮𝘀𝗲: 𝗦𝘆𝗰𝗼𝗽𝗵𝗮𝗻𝗰𝘆 𝗮𝗻𝗱 𝘁𝗵𝗲 𝗱𝗲𝘀𝗶𝗴𝗻 𝗼𝗳 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗺𝗼𝗱𝗲𝗹𝘀 First in a new series 👇 #LawZero #LLM #Sycophancy #SafeAI #ScientistAI

LawZero | Trained to please: Sycophancy and the design of language models

This is the first of a series of posts on problems that derive from the way modern AI systems are built. We start in this post with sycophancy, the tendency of models to flatter or agree with users at...

lawzero.org

𝗡𝗢𝗨𝗩𝗘𝗔𝗨 𝗕𝗟𝗢𝗚𝗨𝗘 Entrainés à plaire: la flagornerie et la conception des modèles de langage Premier article d'une nouvelle série 👇 #LoiZéro #IAResponsable

LoiZéro | Entraînés à plaire : la flagornerie et la conception des modèles de langage

Voici le premier d’une série d’articles consacré aux problèmes découlant de la conception des systèmes d’IA modernes. Dans cet article, nous abordons « la flagornerie », c’est-à-dire la tendance des m...

lawzero.org

Première participation de LoiZéro à ICML 2026, à Séoul. Notre équipe de recherche sera au kiosque B112 toute la semaine. Si vous y êtes, venez nous rencontrer. 👇 6-11 juillet ❘ Coex, Séoul #LoiZéro #ICML2026 #IA @icmlconf.bsky.social

LoiZéro | LoiZéro participera à ICML 2026 à Séoul, élargissant ainsi sa présence mondiale dans le domaine de la recherche sur la sécurité de l'IA

Les chercheurs de l'organisme à but non lucratif fondé par Yoshua Bengio, lauréat du prix A.M. Turing 2018, iront à la rencontre de la communauté internationale de l'IA et rencontreront les meilleurs ...

lawzero.org

LawZero will exhibit for the first time at ICML 2026 in Seoul. Our research team will be at booth B112 all week. We're building safe-by-design AI as a global public good. If you’ll be at ICML, come meet us. 👇 July 6-11 ❘ Coex, Seoul #LawZero #SafeAI #ICML2026 @icmlconf.bsky.social

LawZero | LawZero to Attend ICML 2026 in Seoul, Expanding its Global Safe AI Research Presence

Researchers from the nonprofit founded by Yoshua Bengio, 2018 A.M. Turing Award recipient, will connect with the international AI community and recruit top talent committed to safe-by-design AI.

lawzero.org

La nouvelle stratégie canadienne en matière d’intelligence artificielle constitue une avancée importante, non seulement pour l’innovation, mais aussi pour la sécurité, la gouvernance et la confiance du public. www.pm.gc.ca/fr/nouvelles...

Le premier ministre Carney lance L’IA pour tous, la nouvelle stratégie nationale du Canada en matière d’intelligence artificielle

Aujourd’hui, le premier ministre Mark Carney a lancé L’IA pour tous, la nouvelle stratégie nationale du Canada en matière d’IA. Au cours des cinq prochaines années, cette stratégie donnera lieu à de n...

pm.gc.ca

LawZero - LoiZéro @law-zero.bsky.social · 3mo ago

Canada’s new AI Strategy is an important step forward not only for innovation but also for safety, governance and public trust. www.pm.gc.ca/en/news/news...

Prime Minister Carney launches AI for All: Canada’s new national artificial intelligence strategy

Today, the Prime Minister, Mark Carney, launched AI for All, Canada’s new national AI strategy. Over the next five years, this strategy will introduce new legislation, investments, and programs that e...

pm.gc.ca

𝗡𝗘𝗪 𝗣𝗔𝗣𝗘𝗥:Language models recognize dropout and Gaussian noise applied to their activations. The team introduced an a-semantic perturbation into a language model. See what happened ➡️ lawzero.org/en/publicati... #AISafety #MLSky #LLM #LawZero @yoshuabengio.bsky.social

LawZero | Language Models Recognize Dropout and Gaussian Noise Applied to Their Activations

We provide evidence that language models can detect, localize and, to a certain degree, verbalize the difference between perturbations applied to their activations. More precisely, we either (a) mask ...

lawzero.org

LawZero - LoiZéro @law-zero.bsky.social · 4mo ago

"Intelligence gives power." 💡 Jannis Brühl of Süddeutsche Zeitung traces the career of our Scientific Director @yoshuabengio.bsky.social, and explains how our Scientist AI project represents a completely new approach to developing safe and highly-capable AI. 🇩🇪 www.sueddeutsche.de/wirtschaft/y...

LawZero - LoiZéro @law-zero.bsky.social · 5mo ago

We are pleased to announce the appointment of the Rt. Hon. Dame Jacinda Ardern, former Prime Minister of New Zealand, to our Global Advisory Council. Her appointment is an exciting step to support our mission to develop safe-by-design, highly capable AI systems.

LawZero - LoiZéro @law-zero.bsky.social · 5mo ago

Nous sommes ravis d'annoncer la nomination de la très honorable Dame Jacinda Ardern, ex-première ministre de Nouvelle-Zélande, au sein de notre Conseil consultatif mondial.

Yoshua Bengio @yoshuabengio.bsky.social · 5mo ago

I am deeply honored to have been elected as Co-Chair of the United Nations' Independent International Scientific Panel on AI alongside Nobel Peace Prize laureate @mariaressa.bsky.social.

United Nations Office for Digital and Emerging Technologies@unodet.bsky.social · 5mo ago

At its inaugural meeting, the UN’s Independent International Scientific Panel on AI elected its first Co-Chairs: @yoshuabengio.bsky.social 🇨🇦 @mariaressa.bsky.social 🇵🇭 They’ll guide the Panel’s work, uphold its scientific integrity & report on the Panel’s work twice a year. First report: July 2026

LoiZéro salue cette initiative des gouvernements du Canada et de l'Allemagne ainsi que leur engagement envers un développement sécuritaire de l'IA. 👏 Nous sommes ravis de collaborer avec les deux gouvernements sur cette question importante. www.canada.ca/fr/innovatio...

Le Canada et l’Allemagne signent une déclaration d’intention commune en matière d’intelligence artificielle et lancent l’Alliance pour la souveraineté technologique - Canada.ca

Le ministre de l’Intelligence artificielle et de l’Innovation numérique, l’honorable Evan Solomon, et le ministre de la Transformation numérique et de la Modernisation du gouvernement de l’Allemagne, ...

canada.ca

LawZero applauds this initiative by the governments of Canada and Germany and their commitment to developing safe AI. 👏 We look forward to collaborating with both governments on this important issue. www.canada.ca/en/innovatio...

Canada and Germany sign AI joint declaration and launch Sovereign Technology Alliance - Canada.ca

Today, the Honourable Evan Solomon, Minister of Artificial Intelligence and Digital Innovation, and Karsten Wildberger, Germany’s Minister for Digital Transformation and Government Modernization, sign...

canada.ca

Congratulations to our Scientific Director @yoshuabengio.bsky.social for his appointment to the UN's Independent International Scientific Panel on AI, an important step to build evidence-based foundations for strong global governance in AI.

I am honored to have been appointed to the UN' International Independent Scientific Panel on AI. I look forward to working alongside global experts to advance scientific understanding and ensure that international deliberations are informed by the best available evidence. www.un.org/independent-...

Félicitations à notre directeur scientifique @yoshuabengio.bsky.social pour sa nomination au sein du Panel scientifique international indépendant sur l'IA de l’ONU, une étape importante pour s’assurer que la gouvernance mondiale de l'IA repose sur des bases scientifiques solides.

Nous proposons de repenser l'IA de pointe pour qu’elle soit à la fois performante et sécuritaire dès la conception. Voici un premier billet de blog décrivant certains des objectifs et des composantes de notre projet d’IA-chercheur. lawzero.org/en/unlisted/...

LawZero | The Scientist AI: Safe by Design, by Not Desiring

Scientific theories aspire to describe what is, as opposed to prescribe what ought to be. At LawZero, we take this idea as a design principle for safe artificial intelligence: that understanding—even ...

lawzero.org

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)

🚨 Le Rapport International sur la sécurité de l'IA, présidé par notre directeur scientifique @yoshuabengio.bsky.social, est maintenant disponible !

Today we’re releasing the International AI Safety Report 2026: the most comprehensive evidence-based assessment of AI capabilities, emerging risks, and safety measures to date. 🧵 (1/19)

🚨 The International AI Safety Report, chaired by our Founder and Scientific Director @yoshuabengio.bsky.social, is now available!