Researchers at the Purdue University Polytechnic Institute have developed a new AI software that searches storage sites like Google Drive and DropBox for digital criminal evidence. Data in the cloud can be difficult for law enforcement to access as is identifying the owner of the data. Researchers created a forensic cloud model to scan images, videos, and other files. The AI software flags these files for potential evidence of criminal activity. Purdue University Polytechnic Institute, one of the top institutions for cybersecurity in the country developed the program with StegnoCloud technology. The software is programmed to identify evidence of drug trafficking, child exploitation, illegal firearms transactions, and more. The evidence is complied into a forensic evidence collection system. The project was funded by the National Institute of Justice.
This program, called FileTSAR allows companies offering cloud data storage system to collect logs of flagged data and block accounts containing the data. A report generated by the software can be sent to law enforcement if a search warrant is issued to the cloud provider. The software also reduces evidence storage size and reduces time spent on filtering out false-positive results. Researchers assigned to the project said reducing storatge size and filtering false positives allows for easy transmission of data evidence from the cloud storage providers to law enforcement evidence collection and analysis tools.
The research team at Purdue tested more than 4,500 images and FileTSAR was able to classify criminal activity in the images with a rate of 96% accuracy.
FileTSAR is free for law enforcement agencies. Software developers hope that FileTSAR will allow improved communications among law enforcement agencies at the state, local, national, and even global levels when combating cyber crimes and collecting digital evidence.
“Our new toolkit allows investigators to retrieve network traffic, maintain its integrity throughout the investigation, and store the evidence for future use,” said Seunghee Lee, a graduate research assistant who has worked on the project from the beginning. “We have online videos available so law enforcement agents can learn the system remotely.”
