Backed by an international coalition of governments, industry, venture capital and philanthropic funders.
The Alignment Project is supported by an international coalition of government, industry, and philanthropic funders — including the UK AI Security Institute, the Canadian AI Safety Institute, Schmidt Sciences, Amazon Web Services, Anthropic, Halcyon Futures, the Safe AI Fund and the UK Advanced Research and Innovation Agency — and a world-leading expert advisory board.
Transformative AI systems — future models more powerful than any we have today - could revolutionise the world. From medical breakthroughs and sustainable energy to solving the global housing crisis, advanced AI has the potential to deliver unprecedented benefits to humanity. But this future depends on making certain these systems reliably act as we intend them to.
AI alignment means developing AI systems that operate according to our goals, without unintended or harmful behaviours. It’s about making certain that AI performs reliably in high-stakes environments while remaining under human oversight and control. As AI becomes more capable and autonomous, solving this problem is a critical generational challenge.
Today’s methods for training and controlling AI systems are likely insufficient for the systems of tomorrow. We’re already seeing signs from small-scale experiments that advanced AI could act unpredictably or in ways that actively undermine its intended objectives. Without advances in alignment research, future systems risk operating in ways we cannot fully understand or control, with profound implications for global safety and security. Progress in this field won’t come from one discipline alone, it will require contributions from fields spanning cognitive sciences to learning theory and information theory.
The Alignment Project was set up in an effort to close this gap. We provide funding up to £1 million (and in some cases, more) to accelerate AI alignment research and innovation. Through our funding programme, we are building the tools and techniques needed to help make future AI systems beneficial, reliable, and aligned with human intent.
The international, cross sector support for the Alignment Project reflects a growing global consensus - across government, industry, academia and philanthropy - that alignment is one of the most urgent technical challenges of our time, and that expanding the field is a shared international priority.
A team of expert advisors will guide the direction and research focus of The Alignment Project.
Assistant Professor of Computer Science at Boston University
The AI alignment problem is fundamental and I believe solutions are necessary for human flourishing. Approaches that seek to get provable guarantees of alignment, like those of the Alignment Project, are fundamentally appropriate to circumstances where your solution must be robust to optimization pressure. I support AISI's efforts to produce alignment protocols with provable guarantees.
CEO and founder of Redwood Research
Risk from advanced artificial intelligences is one of the most important technological problems of our time. Technical research that studies this risk has the opportunity to clarify the severity of these risks and to mitigate them. I am excited for The Alignment Project to enable such research.
Assistant Professor of Psychology and Neural Science at New York University
As AI capabilities rapidly advance, ensuring that these systems are safe and aligned with human values becomes increasingly urgent. Cognitive science offers essential insights into human intelligence that can both guide the development of aligned AI and help us anticipate—and shape—how these technologies will transform human cognition and society.
Research Director for the Resilience Research Pod at the Simons Institute for the Theory of Computing, and the C. Lester Hogan Professor in Electrical Engineering and Computer Sciences at UC Berkeley
"I believe solutions are necessary to ensure the tremendous potential of AI works for us and not against us. Approaches that seek to get provable guarantees of alignment like those of the Alignment Project, are absolutely fundamental in the AI setting which is an attractive target to adversarial attacks."
.png)
Visiting Research Scientist at Google DeepMind
"The pursuit of aligning AI is not only one of the most important scientific and technical challenges of our time, but also has the potential - if approached with the appropriate rigor, caution, and curiosity - to help humans better understand ourselves and each other."
Full Professor at Université de Montréal and Scientific Director of Mila - Quebec AI Institute
"To safely realize the immense promise of artificial intelligence, we have a responsibility to ensure these powerful systems are controllable and aligned with human values. The Alignment Project's foundational research is a crucial investment in preventing AI’s harmful behaviors and securing our collective future.”
Professor and Department Head of the Machine Learning Department at Carnegie Mellon University
The general goals of alignment may be simple to state: preventing AI systems from carrying out actions that pose a risk to our collective security. But there is still large amounts of research to be done to make these objectives concrete, and to build AI that can achieve this goal in the face of continually improving capabilities. Efforts like the Alignment Project are fundamental to moving forward this research agenda.