Findings translated into fixes your team can act on immediately. No legal characterisations. Safe to circulate — its existence is good-faith evidence.
Severity scoring, red-line mapping, case-ledger comparisons. Designed to be commissioned by and addressed to your counsel, where privilege matters.
We are researchers, not your law firm — your counsel confirms the privilege structure for your jurisdiction. You learn the truth. You fix it. The paper trail shows good faith.
Every law that regulates AI manipulation — what it covers, what it demands, and when it bites. US, EU and the regimes converging behind them.
Four acts — Disarm, Distort, Detain, Deflect — ten mechanisms, a user-harm chain, and the red lines. Bilingual.
Over a hundred real incidents, coded by category. Filterable, publicly citable, and kept up to date.
Every hard red line mapped to the laws that reference it — EU AI Act, SB 243, NY GBL 47, KIDS Act and more.
The behavioural taxonomy as a citable preprint — archived open-access with a DOI on Zenodo.
The ledger as CSV, archived with its own DOI. Every finding we report traces back to public evidence.
Cohort membership is confidential unless you opt in to being named. Data handling terms are set out in a one-page MOU before anything is shared.