A registry of strategically offensive methodologies to destabilize AI-driven frameworks.
Recent research has explored how to integrate image-poisoning scripts directly into static website build pipelines to protect digital content from unauthorized generative AI scrapers. 3. Context & Related Groups
They define algorithmic sabotage not necessarily as destruction, but as . This can take several forms: algorithmic sabotage research group %28asrg%29
As automation integrates further into healthcare, housing allocation, climate modeling, and warfare, the scope of algorithmic sabotage is expected to widen. The Algorithmic Sabotage Research Group positions itself not as an opponent of technology, but as an opponent of technological totalitarianism.
It is crucial to distinguish the ASRG from the mainstream field of AI safety research. While both fields are concerned with the potential harms of AI, their approaches and goals are diametrically opposed. Context & Related Groups They define algorithmic sabotage
The General in charge slid a folder across the table. “Dr. Vance. We need you to sabotage our own algorithm. Before it does something we can’t take back.”
The is a critical research collective and artistic-academic initiative focused on investigating the intersections of algorithms, power, and resistance. The group is best known for developing the concept of "Algorithmic Sabotage"—a framework for understanding how individuals and groups can deliberately disrupt, confuse, or subvert automated decision-making systems to protest bias, surveillance, and opaque governance. It is crucial to distinguish the ASRG from
To combat the threats of algorithmic sabotage, the ASRG employs a multi-faceted approach:
The ASRG’s tactics have gained attention from prominent figures in the tech world. On August 11, 2025, the blog of pioneering web developer JWZ featured a post detailing the group’s list of “strategically offensive methodologies and purposefully orchestrated tactics.”. The post acknowledges the difficulty of proving the effectiveness of data poisoning but notes that the only people who can confirm it are “The Adversary” themselves—the AI companies whose models have been corrupted.
Creating "tarpits" for AI crawlers that trap them in slow-loading websites filled with "garbage" or fake texts to waste compute time.