Because much of the ASRG’s work is considered pre‑disclosure risk (publishing the method could enable real-world sabotage), few full papers enter the public domain. However, three experiments have been partially declassified by the group’s ethics board:
While the collective is somewhat fluid in its membership, several key projects and conceptual frameworks define their public output: algorithmic sabotage research group asrg
. This document serves as a roadmap for "militant algorithmic agency" and includes several key principles: The First Step is Political: Because much of the ASRG’s work is considered
"We aren't trying to break one model," reads an ASRG internal memo obtained by this journalist. "We are trying to collapse the trust gradient of all open-source weights. If you don't know whether a dataset contains our samples, you cannot safely train a model." you cannot safely train a model."