Forget typosquatting; slopsquatting is the software supply chain threat created by AI coding tools

Slopsquatting is a novel supply chain attack that leverages hallucinations from large language models (LLMs) to introduce malicious code into software development workflows. Unlike traditional typosquatting, which relies on registering misspelled versions of popular packages, slopsquatting exploits AI-generated fictitious package names that appear plausible but do not exist. Attackers register these fabricated package names and populate them with malware, which developers may inadvertently include when following AI coding suggestions.

AI coding assistants often hallucinate package names during code generation, producing recommendations that sound legitimate but are entirely invented. While typosquatting protections exist for common misspellings, they do not cover these AI-generated names, leaving a gap in security. This allows attackers to distribute malicious packages that can remain undetected for extended periods, potentially compromising numerous development environments.

Research indicates that vulnerabilities in open-source packages are increasing rapidly, with an annual growth rate of 98%, and the average lifespan of these vulnerabilities is also rising. This trend underscores the growing security challenges in software supply chains, exacerbated by AI hallucinations. Malicious actors can exploit the predictable nature of hallucinated package names to register harmful packages that deceive developers and evade detection.

LLMs generate responses based on statistical likelihood rather than guaranteed accuracy, resulting in hallucination rates ranging from 23% to over 80%, depending on the model and prompting techniques. Proprietary models tend to hallucinate less frequently than open-source ones, but all remain vulnerable. Adversarial attacks may further increase hallucination rates by manipulating model inputs.

As AI-assisted coding becomes more prevalent, with over 40% of code commits involving AI tools and 72% of users employing AI daily, the risk surface for slopsquatting expands. Developers and organizations must implement verification processes, such as automated checks against official package registries, to identify and block hallucinated packages before they enter production. Continuous monitoring and updated threat intelligence are also critical to mitigating this emerging threat.

Key topics in this update include forget typosquatting slopsquatting, software supply chain threat created, and software supply chain.