A US telecommunications firm, Securus Technologies, has developed an artificial intelligence model trained on years of recorded phone and video conversations involving inmates. This innovative tool is currently undergoing pilot testing to enhance the monitoring of inmate communications, with the goal of identifying and preventing potential criminal activities. According to Kevin Elder, president of Securus Technologies, the company utilized a vast database of inmate calls, including seven years of recordings from Texas prisons, to create AI models capable of detecting suspicious behaviors and conversations. The technology is designed to monitor real-time communications, though Securus has not disclosed the specific locations where these tests are being conducted. Their clientele includes various detention facilities, such as jails housing pre-trial detainees and immigration detention centers.
The AI’s capabilities allow it to analyze phone calls, video interactions, text messages, and emails, flagging notable segments for human review. This process aims to catch criminal plotting at an earlier stage, potentially disrupting serious offenses like human trafficking and gang operations. However, the implementation of such surveillance systems raises significant ethical questions. Bianca Tylek, executive director of the advocacy group Worth Rises, criticized the process as a form of coercive consent, where inmates are not fully informed that their conversations would contribute to AI training. Furthermore, concerns have been raised about Securus’s history of civil liberties violations, including improper recordings of attorney-client conversations.
In response to the ongoing ethical debates, Securus maintains that its AI tool is intended to enhance the efficiency of monitoring rather than target specific individuals. The company also recently achieved a favorable outcome regarding regulatory frameworks that govern how telecom companies can utilize funds collected from inmate calls. Following a Federal Communications Commission (FCC) reform that limited the costs passed onto inmates, Securus lobbied for a reconsideration of these rules, asserting that such changes could impede their ability to maintain effective monitoring. The FCC has since indicated a willingness to support telecom firms in funding their AI initiatives, raising further questions about the balance between public safety and the rights of incarcerated individuals.
Source: An AI model trained on prison phone calls now looks for planned crimes in those calls via MIT Technology Review