The big unknowns about agentic AI
It's my job to think about agentic AI all day and these are the ten questions that I'm grappling with. Thinking about it too? Get in touch
Is this what assessing risk *actually* looks like?
Regulators have spent years trying to get platforms to anticipate harm before it happens. Anthropic’s Mythos release suggests some AI labs may already be adopting similar principles.
How to operationalise platform policy at scale
T&S Insider’s practical, step-by-step guide to making policies consistently enforceable by human reviewers, machine learning models, or LLMs — or a combination of all three
T&S is political. Fund it like it is.
The Trust & Safety Summit reminded us that, if T&S is central to how platforms govern speech, behaviour and risk, it should be treated as a strategic function rather than a cost centre.