Algorithmic Sabotage Research Group %28asrg%29 !free! < 2026 Update >

Rather than attempting to "fix" or optimize existing commercial machine learning models, ASRG explicitly rejects the foundational premises of mainstream AI development. The group’s philosophy shifts away from standard corporate "red teaming"—which it argues functions as free labor used to improve proprietary corporate tech—toward ideological, structural subversion. 1. Countering Algorithmic Authoritarianism

The is an ongoing, conspiratorial, aesthetico-political, practice-led research initiative exploring the intersection of digital culture and information technology. Operating at the volatile intersection of tactical media, digital activism, and critical theory, the group attempts to conceptualize and deploy "algorithmic sabotage" as a legitimate counterweight to modern corporate and state technologies. Rather than attempting to fix or refine biased AI models, ASRG advocates for intentional disruption. They position their work against structural injustices, algorithmic authoritarianism, and the pervasive harms of unrestrained technosolutionism.

While the ASRG is a distinct political-artistic project, it sits within a much wider ecosystem of adversarial AI research, data poisoning, and direct action against web scraping. The summer of 2024 marked an inflection point in the fight against aggressive web crawlers. Anthropic’s ClaudeBot was accused of hammering websites a million or more times a day, and Reddit’s CEO publicly called out AI companies whose crawlers he described as “a pain in the ass to block.” algorithmic sabotage research group %28asrg%29

This philosophy was on full display at major hacker events. According to a 2025 report, while a group called the "Algorithmic Resistance Research Group (ARRG)" (a likely sibling effort) was actively presenting at DEFCON's AI Village, thousands of hackers were convening in Las Vegas to test the boundaries of various Large Language Models. This underscores that the once-fringe idea of "hacking back" against AI is becoming a more mainstream activity within the technical community.

Upcoming like AMRO where they present

: They promote "prefigurative techno-political strategies," often using art as a vehicle for resistance. Key Research and Tactics

Data poisoning relies on feeding generative AI models altered training data. The data looks completely ordinary to a human reviewer but severely degrades an algorithm’s learning process. Rather than attempting to "fix" or optimize existing

To contaminate datasets utilized to train machine learning architectures.

The acronym is common in the tech and security space. You may also be interested in: Drop #17. Manifesto On Algorithmic Sabotage and critical theory

The battle over data is only intensifying. As more large language models train on internet-scale data, resistance movements increasingly see every poisoned data point and trapped crawler as a small victory in a larger war of attrition. The central question facing the ASRG and similar movements is whether sabotage can scale: whether a distributed network of activists, artists, and independent webmasters can meaningfully degrade AI systems reliant on massive data extraction.