A recent test of the "Antiplagiat.VUZ" system revealed that it failed to detect AI-generated text from three different LLMs, rating them as over 98% original. This contradicts claims made by the "Antiplagiat" company that an updated system could detect AI with 98% accuracy. The article explains that common plagiarism detection methods, like shingling and vector similarity, can be easily fooled by AI or misinterpret human writing. A Tatarstan court recently ruled that a student's expulsion based solely on "Antiplagiat.VUZ" results was unlawful, highlighting the unreliability of such tools. AI
IMPACT Undermines confidence in AI detection tools, potentially impacting academic integrity policies and the development of more robust detection methods.
RANK_REASON Article discusses the limitations and unreliability of a specific AI detection tool, including a court ruling against its sole use.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
