Steven Reiss. Thinker · Researcher · Writer · Advisor

Governance

Most of what goes wrong with AI will look like an institution failing slowly.

I advise public institutions and senior decision-makers across Europe and North America on AI adoption and the institutional change it brings. In practice that means the EU AI Act and the governance debate, and a corpus of regulatory, policy, and institutional texts I read with my own tools.

I come to it as an art historian and cultural researcher, after a decade inside the universities, museums, and archives that decide what knowledge survives. The risks I work on are the structural ones the misalignment frame underweights: epistemic erosion, model collapse, the concentration of interpretive power. The alignment work matters, and these sit underneath it. I think the people who study how institutions hold knowledge belong where the rules get written.

Lines of work

The work keeps returning to the same questions.

01Provenance & evidence

What still counts as proof once anything can be generated. AI dissolves the origin of what it makes, and governance has to decide what a society will still accept as evidence.

02Memory & interpretive power

What these systems remember, and what they let us forget. A training corpus is a cultural-memory institution now, and what goes into it is a governance question long before it is a copyright one.

03Adoption & institutions

How institutions absorb a contested technology, and why most adoption stalls. The failure is rarely the technology itself, more often the organization around it.

What I can do

I work as a writer and an advisor. In practice, that takes a few forms:

  • Policy and position writing. Research-grade analysis on AI governance and the EU AI Act, in English and German, under my name or an organisation’s.
  • Briefings and decision papers. The regulation and the risks at the length a leadership meeting actually reads, written so a board can act on it.
  • Orientation. I help a company or an institution work out where AI belongs in what they do, and where it does not. Often the most useful outcome is ruling things out early.
  • Adoption and change. Why AI projects fail inside institutions, and what makes them hold.
  • Talks and teaching. Lectures for leadership and for the public, from executive briefings to museum stages.

Why adoption fails inside institutions, as an interactive investigation: Who Killed the AI Project?

If any of this is useful to you, I am easy to reach.

I answer my own mail.

stenreiss@gmail.com