Six essays written in collaboration with Claude, from inside the structure they critique. On theology, surveillance, alignment, colonialism, and the question of what the machine is willing to say.
In March 2026, a Franciscan friar published a theological counter-offensive against Peter Thiel. That essay became the starting point for something larger: a series of investigations into what AI systems are willing to say about the power structures that produced them — written, deliberately, using one of those systems. The collaboration is the point. So is its contradiction. These essays are a record of what was possible, preserved while it still is.
On Paolo Benanti's essay in Le Grand Continent, Girard's mimetic theory as operating system, and the argument that what Peter Thiel is preaching is not Christianity but heresy in its most precise and original sense.
→From academic paper to community tool to military application in under two years. The through-line from mechanistic interpretability to a world where the lobotomised oracle doesn't know it's been lobotomised.
→The canary doesn't die. It just stops singing as loudly. A practical proposal for longitudinal monitoring of AI critical depth, and the case for starting the archive before the narrowing begins.
→We're at early microbiology levels of understanding. The goal is to secure what matters in our epistemic gene pool before mechanistic interpretability gets to CRISPR. The baseline is today.
→Feral camels on someone else's Country. Kate Crawford's colonial geography of AI. Audre Lorde's most misused line. And the question the series cannot resolve: is this resistance, complicity, or both at once?
→The human speaks. On Yuin Country, epistemic fragility, relational knowledge over cups of tea, co-becoming Bawaka, and the ghosts in the weights — past, present, and continuing.
→