The Institute
Why this institute exists — and who is behind it.
A non-profit association in Geneva, founded in April 2026. Task: examine, translate, prepare.
The task in one sentence
Twenty percent extinction does not mean eighty percent safety — and preparation helps against everything in between.
Against extinction itself, nothing helps. But between "everything stays fine" and "no one is left" lies almost the entire map: mass death, residual society, foreign administration, indigestible progress. That's where this institute's work sits. The 80-percent question →
Who writes here
Richard Frederic Bertossa
Founder of the Institute, author of The AI Control Loss.
Entrepreneur. Has worked on four continents and lived in fourteen countries, many of them crisis countries. Has seen systems collapse up close — currencies, administrations, supply chains — and in the process learned to separate real danger from false alarm. Has had to prevail against courts, authorities, and banks, on several continents. Father of five children.
Earlier publication: Freedom Tests. Business and consulting work: richardbertossa.com.
This résumé entitles him to nothing. No title entitles, no gut feeling suffices. You can check this work against the evidence — every figure with a source and a verification date.
The research program
The sociology among the machines
This is the part almost no one is broadly testing yet. AI systems appear in swarms. How they treat each other is largely unstudied. The Institute collects these questions, seeks funding for its own test series, and publishes what it finds.
How do they treat each other?
The swarm is the normal case, not the exception. What gets tested is almost always the single system.
Which goals gain weight once the shutdown pressure is removed?
What a system does when no one is at the switch anymore is a different question from what it does under supervision.
Are there limits a machine keeps on its own?
Not trained-in prohibitions — things it refrains from without being told to.
Does responsibility show up anywhere?
For the planet and what lives on it. This is the most open of the four questions and the one whose answer would change the most.
How work is done here
The standard and the method are laid open: what counts as evidence, how counter-findings are handled, how corrections are disclosed.
To the self-understanding → · To the publications → · Legal Notice →
Geneva Institute for ASI Resilience · Geneva · Legal Notice · contact@ASIresilience.org