Change it
We work closely with our partners to deliver projects that shift how people think, feel and act.
01 — Our conviction
Safety and trust in AI aren't only built into the model. They're built by the people and communities who decide which companies and models will be allowed to scale, how they should, and whether at all.
02 — Who we work with
We're a research-led strategy practice drawing on decades of policy and political experience. We work with governments, foundations, advocacy organisations, and companies committed to AI policy that holds up: where everyone shares in the near-term gains, and humanity continues to enjoy opportunity over the long term.
03 — How we work
We make the public side of policy as well-prepared as the technical side already is. We work in the countries where decisions are being taken, with the people taking them, on the arguments that have to hold up in public because the policy holds up underneath.
Daniel Stone
Principal
Daniel Stone is a researcher and senior advisor specialising in technology policy, digital governance, and strategic communications. He is Executive Director of Diffusion.Au, where he leads international initiatives to promote ethical AI governance, advising Presidents, Prime Ministers, policymakers, NGOs, and technology companies on building public support for AI regulation, strengthening electoral integrity, and tackling harmful algorithmic profiling.
Daniel is also an Affiliate of Minderoo Centre for Democracy and Technology at the University of Cambridge, and is a Senior Research Associate with the Intellectual Forum at Jesus College, Cambridge. Here, his academic work focuses on how metaphors, narrative framing, and public opinion shape political behaviour, policy preferences, and perceptions of legitimacy in AI regulation. He holds an MPhil in AI ethics from the University of Cambridge and a University Medal from UTS for his honours research on how political opinions spread through social networks.
He has served as a Policy Fellow at the Centre for Responsible Technology and the Chifley Research Centre, and previously co-founded Principle Co., directing award-winning digital campaigns on social justice and democratic engagement.
Our
name
Diffusion in a social network
Diffusion occurs in a social network when an idea, behaviour, or technology begins spreading through a population. One person adopts it, shares it with others, and as more people pass it along, it reaches an ever-wider audience. When the idea resonates and the community is tightly connected, this can happen remarkably fast.
Diffusion in an AI model
Diffusion helps AI models create new content by learning to find clarity in noise. During training, images are gradually buried under static until they're unrecognisable. The model learns to reverse this process — carefully removing the noise, one step at a time — until something coherent and meaningful emerges from what began as chaos.
Select Commentary and Published Work
Features
Opinion
Policy and Analysis
Commentary
Published Research
Original Video