MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202641085130 A) filed by Srm University- Ap on July 11, 2026, for System And Method For Detecting Hallucinated Information In Ai- Generated Summaries.
Inventors include Dr. Shaik Rafi; Kolla, Devasish Viswanadh; and Dr. Ranjita Das.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: A system (100) for detecting hallucinated information in AI-generated summaries comprises a processor (102) and a memory (104) storing instructions that, when executed by the processor (102), cause the system (100) to implement a similarity evaluation module (106) configured to compute similarity metrics between a generated summary and reference content, a natural language inference module (108) configured to determine a logical relationship between a source document and the generated summary, a sentence analysis module (110) configured to evaluate individual sentences of the generated summary against the source document, and a risk computation module (112) configured to compute a composite faithfulness risk score based on outputs from the similarity evaluation module (106), the natural language inference module (108), and the sentence analysis module (110). The system (100) processes the generated summary through a multi-stage verification pipeline, classifies the generated summary into a reliability category based on the composite faithfulness risk score. (FIG. 1)
Disclaimer: Curated by HT Syndication.
