
A new security threat known as 'slopsquatting' is targeting developers who rely on AI assistants for coding tasks. Because LLMs frequently hallucinate non-existent software packages, attackers are proactively registering these fake names on public registries like PyPI. When a developer follows an AI's suggestion to install a non-existent library, they inadvertently download malicious code. Research presented at USENIX Security 2025 revealed that nearly 20% of AI-suggested packages are hallucinations, with many being generated predictably across different models. This allows attackers to identify and squat on these names effectively. Experts warn that traditional typosquatting defenses are insufficient because these names are often not simple misspellings. To mitigate this risk, developers are advised to treat all AI-generated package names as untrusted input, verify their existence before installation, and implement strict dependency management practices, such as using lockfiles and private registries, to prevent automated agents from pulling malicious dependencies.
This is a summary. Read the full article at the original source:
Dev.toRelated stories
The Russian-linked advanced persistent threat (APT) group known as Star Blizzard has shifted its tactical approach in recent phishing campaigns. Accor…
In this article, Yandex information security expert Aidar Sabirov describes a technical solution for complying with legal requirements that mandate au…
Cloudflare has announced plans to provide quantum-safe TLS certificates, positioning itself as a leader in preparing internet infrastructure for the p…



