Show HN: Lmscan – Detect AI text and fingerprint which LLM wrote it (zero deps) (github.com)

Lmscan, a new open-source tool, has been launched to detect AI-generated text and identify which large language model (LLM) authored it. This software serves as a free alternative to existing paid solutions like GPTZero and Originality.ai, emphasizing user accessibility by requiring no subscriptions, API keys, or internet connection. Lmscan operates offline and analyzes text using 12 statistical features grounded in computational linguistics, providing insights such as AI probability and stylistic patterns associated with specific LLMs like GPT-4 and Claude. This development is significant for the AI/ML community as it democratizes access to AI detection tools, allowing individuals, educators, and researchers to scan documents for AI-generated content without incurring costs or entering cumbersome contracts. Lmscan's explainable analysis provides detailed reasoning on why a text is likely AI-driven, thereby enhancing understanding of AI writing patterns. While it excels at detecting strong AI signatures and offers per-sentence breakdowns, it has limitations, such as reduced effectiveness on manually edited or short texts and a current focus on English language content. Overall, lmscan represents a crucial step toward transparency and understanding in the evolving landscape of AI-generated communications.