By Harry C. Goodall Jr.
A new artificial intelligence (AI) system targets key words and phrases during monitored phone calls from inside prisons and jails, according to an ABC News article.
Sheriffs and wardens nationwide use AI technology to aid in unsolved crimes such as drug smuggling, attempted suicides and violence in real time.
Legally mandated warnings precede every phone call to inform the patrons their call is being recorded. A vast amountof inmates still reveals incriminating information, according to data given to ABC news by the technology company.
“If I got to stay longer than November… I’m killing all of them when I get out…and I mean it,” said one person during a phone call, according to the article.
Corrections officials in Alabama intercepted a phone call in which a prisoner instructed his wife on how to smuggle Suboxone, an opioid withdrawal aid, into the prison. He told her to first dilute it in water, then use a makeup brush to paint it on postcards to be mailed into the prison. This is just one of many situations in which inmates have made incriminating statements via a previously warned recorded call.
The AI systems in these cases utilized speech-recognition technology, semantic analytics and machine learning software. The system then builds an expanding database of searchable words, part of a global revolution in neural networks, the article stated.
The prison phone system is a $1.2 billion a year industry, according to data by the Prison Policy Initiative. It was also once used to track the locations of people called by the inmates until the U.S. Supreme court ruled that practice illegal.
LEO Technologies is the leader in this innovative technology and has its own investigation division outside of the prison system. This division feeds the databases with keywords, phrases and prison slang based on the region and area of the prison. LEO Technologies then promptly notifies its law enforcement partners if the system picks up phrasing or suspicious language. The technology is near real time and offers rapid response. The company states that it has thwarted dozens of attempted suicides the past two years over several states.
GTL and Securus, the nation’s two largest providers of phone service to prisons and jails, are developing their own call analytics technologies. LEO, which contracts with GTL, is operating in five U.S. states, according to company officials.
Prison pressing priorities are (1) controlling prison contraband, (2) inmate attacks on infrastructure security systems, and (3) unmonitored inmate communication. This data was provided by a National Institute of Justice study of our nation’s nearly 7,000 correctional institutions. The data was published this year by the RAND Corporation, according to the article.
“One of the biggest operational issues that has plagued this industry of automated inmate telephone recording has been the lack of staffing to monitor every single call,” said John Shaffer, a corrections technology expert. “And frankly, most inmate calls are innocuous.”
“There (have) been different approaches over the years,” he said. “I remember Walla Walla penitentiary out in the state of Washington used to have officers sitting there in the towers at night listening to inmate calls for hours and hours – and that was their solution. And it never really worked as an operational opportunity.”
LEO Technologies’ system can range in price, generally between $500,000 and $600,000 a year per 1,000 inmates monitored in the facility in which is installed.
Jails and prisons “are looking at this very seriously, and several are using it – trying to figure out the right balance and mix of human and technology,” said Jonathan Thompson, executive director of the National Sheriff’s Association.
Source: San Quentin News
Go to thecellblock.net and subscribe to The Official Blog of The Cell Block, where we provide you raw, uncensored news, entertainment and resources on the topics of prison and street-culture from a true, insider’s perspective, and follow us on social media @thecellblock and @mikeenemigo.