, a set of ten principles (numbered 0 to 9) designed to turn radical theory into direct praxis. Their philosophy includes: Rejection of "Algorithmic Humiliation"
As AI continues to permeate various sectors, the work of ASRG and similar research groups becomes increasingly critical. Future directions for ASRG include:
While the name sounds like something lifted from a William Gibson novel, the ASRG is a very real, albeit shadowy, coalition of machine learning researchers, digital artists, and adversarial AI specialists. Their mission statement is short and provocative: "To render the unauthorized scraping of creative works for generative AI economically inviable through technical sabotage." algorithmic sabotage research group asrg
Greetings, fellow disruptors!
Documenting historical and contemporary instances where marginalized groups have successfully subverted automated systems. , a set of ten principles (numbered 0
The ASRG did not emerge from a university lab or a corporate R&D department. According to leaked whitepapers and anonymous interviews with founding members (who all insisted on Signal voice calls with voice changers), the group coalesced in late 2022—just weeks before the public explosion of Stable Diffusion and Midjourney.
The group compiles structural tactics aimed at the intentional corruption of data workflows—often referred to historically as throwing a "sabot" (wooden shoe) into the industrial looms. By feeding mislabeled, heavily distorted, or adversarial data streams into public vectors, creators force AI companies to expend heavy manual resources cleaning their data pools. Their mission statement is short and provocative: "To
Historically, the word "sabotage" conjures images of 19th-century industrial workers throwing their wooden shoes ( sabots ) into factory looms to halt exploitative automated production. The ASRG repositions this concept for the 21st century.
The contributions of ASRG to the field of adversarial machine learning have been substantial:
The core theoretical document of the ASRG is the Manifesto on "Algorithmic Sabotage" . Initially published in English, Greek, and German, the manifesto has since prompted an international call for translations, reflecting the group's global ambitions.
For the ASRG, this wasn't a bug—it was a vulnerability.