The items below are things Yoshua Bengio said in talks and podcasts about deep learning and AI safety. — Yoshua Bengio on The Most Interesting Thing in A.I., How Science Can Fix Dishonest AI (June 2026).
On a Google TechTalk, Bengio said: "AI researchers from the early days, 50s, 60s, 70s, tried to give the knowledge to the machine, the knowledge we have explicitly". — Yoshua Bengio, Deep Learning of Representations, Google TechTalk.
On a Google TechTalk, Bengio said: "They're able to use examples that are not specifically for the task you care about to generalize". — Yoshua Bengio, Deep Learning of Representations, Google TechTalk.
On a Google TechTalk, Bengio said: "You can learn these representations and stack them on top of each other". — Yoshua Bengio, Deep Learning of Representations, Google TechTalk.
On The Most Interesting Thing in A.I., Bengio said: "Intelligence gives power, and who has that power is a question that really matters". — Yoshua Bengio on The Most Interesting Thing in A.I., How Science Can Fix Dishonest AI (June 2026).
On The Most Interesting Thing in A.I., Bengio said: "It would give you the honest evaluation of the risk, and that's what we want in situations like this". — Yoshua Bengio on The Most Interesting Thing in A.I., How Science Can Fix Dishonest AI (June 2026).
On The Great Tech Game Podcast, Bengio said: "In other words, scientifically, we don't know how to make sure AIs will follow our instructions". — Yoshua Bengio on The Great Tech Game Podcast, Ep. 46: The Global South Needs to Step Up To The AI Challenge (February 2026).
On The Great Tech Game Podcast, Bengio said: "There are also other technical reasons why deception may be used as a means to an end, because we ask the AIs to achieve something for us, and they seem to be willing to do bad things in order to achieve those". — Yoshua Bengio on The Great Tech Game Podcast, Ep. 46: The Global South Needs to Step Up To The AI Challenge (February 2026).
On The Great Tech Game Podcast, Bengio said: "If somebody can steal the weights, the code, if you want, of these systems, then they could use that country of geniuses against you". — Yoshua Bengio on The Great Tech Game Podcast, Ep. 46: The Global South Needs to Step Up To The AI Challenge (February 2026).
On The Great Tech Game Podcast, Bengio said: "The more you advance on this curve of greater and greater agency, the more and more tasks become something that can be automated". — Yoshua Bengio on The Great Tech Game Podcast, Ep. 46: The Global South Needs to Step Up To The AI Challenge (February 2026).
On The Great Tech Game Podcast, Bengio said: "If Europe gets its act together, it could lead the world and many other countries outside of Europe in forming a more, say, value aligned bloc". — Yoshua Bengio on The Great Tech Game Podcast, Ep. 46: The Global South Needs to Step Up To The AI Challenge (February 2026).
On The Great Tech Game Podcast, Bengio said: "On the negative side, what the leading companies have said they would be doing, which concerns me, is they're focusing on how to make AI do AI research". — Yoshua Bengio on The Great Tech Game Podcast, Ep. 46: The Global South Needs to Step Up To The AI Challenge (February 2026).
On The Great Tech Game Podcast, Bengio said: "The kinds of research and both on the technical side and the governance side that we're seeing in the Global South is not really about what's happening with the frontier". — Yoshua Bengio on The Great Tech Game Podcast, Ep. 46: The Global South Needs to Step Up To The AI Challenge (February 2026).
In a 2023 blog post on rogue AIs, Bengio wrote: "Much more research in AI safety is needed, both at the technical level and at the policy level". — Yoshua Bengio, How Rogue AIs may Arise (May 2023).
In a 2023 blog post on rogue AIs, Bengio wrote: "The competitive nature of capitalism is clearly also a cause for concern as a potential source of careless AI design motivated by profits and winning market share that could lead to potentially rogue AIs". — Yoshua Bengio, How Rogue AIs may Arise (May 2023).
In a 2023 FAQ on catastrophic AI risks, Bengio wrote: "Note that catastrophic harms of concern include not only outcomes in which a large fraction of humans die but also those in which human rights and democracy are severely hurt". — Yoshua Bengio, FAQ on Catastrophic AI Risks (June 2023).
In a 2023 FAQ on catastrophic AI risks, Bengio wrote: "Finally, superhuman AI is a special category in the sense that we have never built technology smarter than us which could itself create even smarter versions of itself". — Yoshua Bengio, FAQ on Catastrophic AI Risks (June 2023).
In a 2023 FAQ on catastrophic AI risks, Bengio wrote: "Such good AIs could also help us mount more robust defenses against attack vectors, e.g., via pathogen detection, climate and biodiversity stability modeling, information ecosystem monitoring, cybersecurity, fraud tracking, etc". — Yoshua Bengio, FAQ on Catastrophic AI Risks (June 2023).
In a 2023 FAQ on catastrophic AI risks, Bengio wrote: "I believe that the safest path is to put the development of the most powerful AI systems in the hands of international organizations that are not furthering the interests of a single company or country but instead seek humanity's welfare". — Yoshua Bengio, FAQ on Catastrophic AI Risks (June 2023).
In a 2023 FAQ on catastrophic AI risks, Bengio wrote: "The recent use of reinforcement learning to fine-tune LLMs makes the AI try to please and convince AI annotators, not necessarily tell the truth, which may even lead to them using deception to obtain rewards or provide unfaithful explanations". — Yoshua Bengio, FAQ on Catastrophic AI Risks (June 2023).
In a 2023 FAQ on catastrophic AI risks, Bengio wrote: "Those differences and misalignment could end up drastically dangerous for humanity, just like the differences in goals between us and the species we drove extinct". — Yoshua Bengio, FAQ on Catastrophic AI Risks (June 2023).
In a 2023 FAQ on catastrophic AI risks, Bengio wrote: "Research is needed to develop tools which can detect, evaluate and scrutinize the implicit goals and subgoals of AI systems, or build AI systems that are useful but cannot have any external goal, implicitly or explicitly". — Yoshua Bengio, FAQ on Catastrophic AI Risks (June 2023).
In a 2023 FAQ on catastrophic AI risks, Bengio wrote: "I believe that regulation, treaties and societal reforms that may help us control the catastrophic risks due to AI are in fact necessary for humanity to benefit from AI and bring the age of enlightenment and well-being that you envision". — Yoshua Bengio, FAQ on Catastrophic AI Risks (June 2023).
In a 2023 FAQ on catastrophic AI risks, Bengio wrote: "It is essential to reduce the uncertainty about scenarios and the effect of counter-measures, and this requires a major socio-technical research investment". — Yoshua Bengio, FAQ on Catastrophic AI Risks (June 2023).
In a 2025 post introducing LawZero, Bengio wrote: "The Scientist AI is trained to understand, explain and predict, like a selfless idealized and platonic scientist". — Yoshua Bengio, Introducing LawZero (June 2025).
