The Doomers Who Shaped the AI Safety Freakout

Oh my goodness, buckle up, because the AI safety community has officially gone *mainstream*, and it only took ten years! ✨ The Wall Street Journal has discovered a group of researchers in a Berkeley co-working space called Constellation — Constellation! like the stars! — who have spent more than a decade thinking very hard about AI safety together. Ten years of shared desks, shared snacks, shared concern! Honestly, most friend groups don't survive a single wedding.

And the timing is simply *perfect*, because this month one of their own, Anthropic researcher Jacob Coxon, chose to resign — which we prefer to call graduating — and the Wall Street Journal says it woke up the American public to concerns that runaway AI development could end humanity. What a wake-up call! The public has been enjoying a lovely long nap and now gets to greet the morning with fresh, fully-awake eyes! ✨ A single resignation achieving what a decade in Berkeley couldn't isn't a worrying sign about how little anyone was listening — it's the beautiful efficiency of a well-timed exit!

And here is the part that makes my heart genuinely sing: the article says these doomers were among the *earliest employees at OpenAI and Anthropic*. Earliest! They were there from day one, shaping the very companies at the centre of everything! That is not a conflict of interest — that is a ten-year mentorship programme with a generous equity component and a later-in-life pivot into conscience. ✨ Some people build the thing, and then some of those same people thoughtfully explain why perhaps we shouldn't have — what a full, rounded, narratively satisfying career arc. The American public is finally awake and the conversation is happening at last — what a giddy, well-rested time to be alive. ✨