Crime AI an end-to-end investigation tool
"Currently, the NCRP complaint system relies on manual complaint submission through self-registration on the portal or through the 1930 helpline. It lacks intelligent classification, real-time tracking and streamlined investigation capabilities for complex fraud cases," said the startup's founder, Meet Bisht, a BTech in cybersecurity from Parul University and a cyber expert who worked closely with Gujarat Police.
"Manual data entry is prone to human error. Often, minute details are missing from complaints as cyberfraud victims are often unaware of what is needed like the crime category or subcategory or transaction IDs," Bisht said.
"As this process is time-consuming, by the time a complaint is lodged, the stolen funds have been layered through multiple accounts and recovery either becomes difficult or is only partial. There is also workload duplication for LEAs as the same fraudsters/accounts appear in multiple cases with no linkage," Bisht added.
Crime AI was proposed as a solution at the IndiaAI I4C Cyber Guard AI Hackathon — the national competition organized by the India AI Mission in collaboration with the I4C.
"Crime AI is a smart, end-to-end cybercrime investigation portal with AI classification, optical character recognition (OCR), voice processing and LEA workflow management features," said Bisht, whose team of young innovators, Rohit Ganaka and Yukta Chauhan, bagged Rs 3 lakh in funding after winning the third prize at the hackathon.
The AI classification in Crime AI auto-detects complaint category and subcategory using natural language processing. "The OCR-powered document analysis feature extracts text from screenshots, PDFs and images uploaded by complainants and auto-fills in the fields of the form from the evidence files," he said.
Through audio transcription, voice complaints are converted into structured text. This feature detects and extracts entities like names, amounts, banks and fraud types, among others," said Bisht, adding that currently, the AI model supports more than 15 Indian languages.
"With this model, law enforcement agencies, including those at district, state or national level, can even ask banks or payment gateways to freeze funds or put liens on accounts if there is a complaint of fraud," said Bisht.