Research.

Original work on how AI governance holds in regulated organisations: one flagship programme, long-form articles, an open thought experiment and working instruments. Financial services is the proving ground; each piece states what generalises.

Flagship programme

Recently published

All articles

Projects

All projects
Responsible AI by design Research programme
Live · Twelve domains that must hold true, and an eight-stage lifecycle that proves it.
Minimum Humans to Run a Bank Thought experiment
Live · Can a small human core govern a UK-regulated retail bank run by AI agents?
ASD-STE100 Simplified Technical English Instrument
Live · Controlled language for LLM output, checked rule by rule.
craft-a-bot Repository
Alpha · An agentic LLM builder and test environment, built like a construction toy.
Algorithmic Impact Assessment Tool Instrument
Live · A working Algorithmic Impact Assessment tool, not a template of one.
Model-Card Repository
Live · One YAML source of truth, four synchronised model card formats.

Contents

  • Articles Long-form analysis, argued and sourced.
  • Thought Experiment Can a small human core govern a UK-regulated retail bank whose work is done by AI agents?
  • Projects Programmes, instruments and public code, each with governance notes.
  • News Regulatory and standards developments, with what each one changes.
  • Learning Ordered paths through the library, scoped to an audience.
  • Industries The library assembled for one sector, starting with financial services.

Axiom Verity is an independent, non-commercial research lab for AI governance, safety, security and compliance. Nothing here is sold or offered as a service. Views are the author's own. Independence statement