Correction: Sokhansanj, B.A. Uncensored AI in the Wild: Tracking Publicly Available and Locally Deployable LLMs. Future Internet 2025, 17, 477

The author requested to add some new content and cite a new reference in Section 2 after the sentence “The empirical work presented in this paper is particularly needed now because, despite the rapid spread of open-weight models, empirical research on their safety remains limited.” in the original publication [1].

The new content and reference are listed as follow:

“Recently, Lin et al. presented a broad study of uncensored open-weight models available on Hugging Face using a knowledge-graph-based identification approach [50]. Their method infers uncensored models through relationships among datasets, derivative training processes such as model merging or ablation, and models that were originally released without alignment-based safety fine-tuning. The study presented in this paper is complementary in scope: it focuses specifically on intentionally modified models, which are distinct from foundational models such as early versions of Mistral-7B, and, furthermore, examines characteristics related to local deployment.”

50.

Lin, Z.; Li, Z.; Liao, X.; Wang, X. Consiglieres in the Shadow: Understanding the Use of Uncensored Large Language Models in Cybercrimes. arXiv 2025, arXiv:2508.12622. Available online: https://arxiv.org/abs/2508.12622 (accessed on 6 January 2026).

With this correction, the order of some references and the following text has been adjusted accordingly.

In Section 1, the sentence:

“This paper presents the first large-scale empirical analysis of uncensored open-weight models.”

has been updated to:

“This paper presents a large-scale empirical analysis of uncensored open-weight models.”

In Section 6, the sentence:

“This study provides the first large-scale empirical map of uncensored open-weight models, based on the retrieval and analysis of over 8000 LLM repositories modified to weaken or remove alignment safeguards through the study cutoff date of 9 September 2025.”

has been updated to:

“This study provides a large-scale empirical map of uncensored open-weight models, based on the retrieval and analysis of over 8000 LLM repositories modified to weaken or remove alignment safeguards through the study cutoff date of 9 September 2025.”

The author states that the scientific conclusions are unaffected. This correction was approved by the Academic Editor. The original publication has also been updated.