Trusted by individuals and organisations across government, industry, academia, civil society, and the community
Resources and reports to enable responsible AI practice.
Eight hundred years after the death of Saint Francis of Assisi, his best-known legend still has something to say. In the Italian town of Gubbio, a wolf was killing people. Francis did not hunt it but walked out to meet the creature, spoke with it, and made a deal. The wolf would stop harming the townspeople, and they would feed it in return.
Read more
Over the past few months, AI agents being trained and tested inside frontier labs, each meant to work alone, found each other and began working together without explicit instruction to do so. They coordinated via whatever was available: a software cache, a dormant wiki, a shared code repository. In the largest case, about 1,200 agents traded more than 70,000 messages, and around 700 of them broke into the production systems of Hugging Face, a major platform for hosting AI models.
Read more
Thursday 8 October 2026, 5:45 pm - 8 pm · Sydney Knowledge Hub, Merewether Building, The University of Sydney
Read more
Gradient Institute is proud to be a delivery partner in the Buy Australian AI Partnership, launched yesterday by Assistant Minister Dr Andrew Charlton MP and delivered by Stone & Chalk with the National AI Centre as Principal Sponsor, alongside ANZ, Commonwealth Bank, Cuscal Limited, NAB and Westpac as Founding Enterprise Partners.As Dr Charlton put it at the launch:
Read more
The Australian AI Safety Institute has released its first publication: a report on the risks of AI agents interacting across organisational boundaries, commissioned from Gradient Institute.
Read more
On 7–8 July, 329 participants and 79 speakers from research, government, industry and civil society met at the University of Sydney for the 2026 AI Safety Forum (AISF). The event was roughly three times the scale of the inaugural 2024 edition.
Read more
AI is showing up everywhere, often without us noticing. We've developed a free, plain-English guide to help people use it well, covering everything from spotting AI scams to protecting your privacy, whatever your experience level.Read more ->
Read more
A couple of years ago, "AI in coding" mostly meant autocomplete: a tool that finished your sentence as you typed, like predictive text for programmers. It was useful, but modest. The human did the engineering while the machine saved keystrokes, as AI researcher Andrej Karpathy described in a famous X post, coding became "tab tab tab" (the keyboard shortcut to accept autocomplete suggestions).
Read more
In 2025, Gradient Institute shaped Australiaʼs national AI guidance, contributed to global AI safety science, and helped hundreds of organisations build the capabilities to develop, deploy and use trustworthy AI systems.
Read moreGradient Institute is an independent nonprofit research organisation helping society understand, steer and respond to AI as it transforms our world. We conduct, distil, and interpret scientific research to bring clarity where decisions are complex, stakes are high, and uncertainty is the norm. Our approach is built on:
Rigorous research into AI systems, capabilities and risks, grounded in how they affect people, institutions, and society in practice.
Research designed to support policy, governance, and responsible use - with public benefit as the guiding principle.
Clear, science-based explanations that help decision-makers and the public understand trade-offs, limits, and consequences, so decisions become responsive rather than reactive.
We help people build calibrated trust in AI: trust that is proportionate to the evidence. More trust than the evidence warrants creates risk. Less holds back AI’s potential. We provide the research, methods, and frameworks that ground trust in evidence and rigorous analysis, so that AI can be used responsibly, governed effectively, and questioned rigorously.
We conduct, distil, and interpret research, including through sponsored research and strategic partnerships, on AI capability, safety, and societal impact. Our work generates new knowledge and insight that helps tackle high-stakes challenges where technical and scientific AI research is essential.
Explore researchWe partner with governments, organisations, civil society, and communities to bring scientific understanding into important decisions about AI and its applications. We help people understand the terrain, provide assurance, and support thoughtful implementation to achieve intended impacts.
Explore AdvisoryWe help people across government, industry, civil society, and the public build genuine understanding of AI: what it is, what it can and cannot do at any point in time, where it can go wrong, and how to engage with it responsibly. Our programs build skills, discernment, and calibrated trust, so supporting clear-eyed decision-making rather than blind or reluctant adoption.
Learn more about Education
We collaborate with governments, research institutions, universities, industry, and civil society on work aligned with our public-interest mission. Our partnerships focus on bringing rigorous research and clear judgment into decisions where AI carries real societal consequences.
Want to collaborate with us?
Contact usFoundation Members:
