ROUGHLY SIX MONTHS ago at New York’s Sing Sing prison, John Dukes says he was brought out with cellmates to meet a corrections counselor. He recalls her giving him a paper with some phrases and offering him a strange choice: He could go up to the phone and utter the phrases that an automated voice would ask him to read, or he could choose not to and lose his phone access altogether.

Dukes did not know why he was being asked to make this decision, but he felt troubled as he heard other men ahead of him speaking into the phone and repeating certain phrases from the sheets the counselors had given them.

“I was contemplating, ‘Should I do it? I don’t want my voice to be on this machine,’” he recalls. “But I still had to contact my family, even though I only had a few months left.”

So when it was his turn, he walked up to the phone, picked up the receiver, and followed a series of automated instructions. “It said, ‘Say this phrase, blah, blah, blah,’ and if you didn’t say it clearly, they would say, ‘Say this phrase again,’ like ‘cat’ or ‘I’m a citizen of the United States of America.’” Dukes said he repeated such phrases for a minute or two. The voice then told him the process was complete.

“Here’s another part of myself that I had to give away again in this prison system,” he remembers thinking as he walked back to the cell.

Dukes, who was released in October, says he was never told about what that procedure was meant to do. But contracting documents for New York’s new prison phone system, obtained by The Appeal in partnership with The Intercept, and follow-up interviews with prison authorities, indicate that Dukes was right to be suspicious: His audio sample was being “enrolled” into a new voice surveillance system.

In New York and other states across the country, authorities are acquiring technology to extract and digitize the voices of incarcerated people into unique biometric signatures, known as voice prints. Prison authorities have quietly enrolled hundreds of thousands of incarcerated people’s voice prints into large-scale biometric databases. Computer algorithms then draw on these databases to identify the voices taking part in a call and to search for other calls in which the voices of interest are detected. Some programs, like New York’s, even analyze the voices of call recipients outside prisons to track which outsiders speak to multiple prisoners regularly.

Corrections officials representing the states of Texas, Florida, and Arkansas, along with Arizona’s Yavapai and Pinal counties; Alachua County, Florida; and Travis County, Texas, also confirmed that they are actively using voice recognition technology today. And a review of contracting documents identified other jurisdictions that have acquired similar voice-print capture capabilities: Connecticut and Georgia state corrections officials have signed contracts for the technology (Connecticut did not respond to repeated interview requests; Georgia declined to answer questions on the matter).

Authorities and prison technology companies say this mass biometric surveillance supports prison security and fraud prevention efforts. But civil liberties advocates argue that the biometric buildup has been neither transparent nor consensual. Some jurisdictions, for example, limit incarcerated people’s phone access if they refuse to enroll in the voice recognition system, while others enroll incarcerated people without their knowledge. Once the data exists, they note, it could potentially be used by other agencies, without any say from the public.

It’s particularly alarming, they add, that the technology’s use in prisons can ensnare people beyond their walls. “Why am I giving up my rights because I’m receiving a call from somebody who has been convicted of a crime?” asks Jerome Greco, a digital forensics attorney at New York’s Legal Aid Society. Greco argues that the mining of outside parties’ voice prints should require a warrant. “If you have a family member convicted of a crime, yet you haven’t been, why are you now having your information being used for government investigations?”

The Spread of Voice Recognition Technology

Voice-print technology works by dissecting physical features that distinguish individuals’ voices, such as their pitch. With this data, the program’s algorithm generates a computer model of their vocal signatures, known as “voice prints,” which can be stored in a database for comparisons with utterances recorded in the future.

In recent years, voice recognition technology has come to be associated with consumer offerings, like Amazon’s Alexa and Apple’s Siri, but the technology was originally developed for military and intelligence applications. Over a decade ago, as The Intercept reported, U.S. intelligence agencies were using voice recognition programs to identify the voices of top Al Qaeda officials in their online audio postings.

Similarly, the algorithms and structure behind the prison telecommunications firm Securus Technologies’ particular voice software, known as Investigator Pro, were developed in part through a $50 million grant from the Department of Defense. The software was licensed to JLG Technologies, a company that Securus acquired in 2014. According to Securus’s 2017 proposal for New York, the technology was developed because “DOD needed to identify terrorist calls out of the millions of calls made to and from the United States every day.”

But it wasn’t long before major prison technology firms, such as Securus and Global Tel Link, began marketing the technology to U.S. jurisdictions that were seeking to extract and store voice prints associated with incarcerated people in their systems. “IPRO [Investigator Pro] has a 10-year track record of providing pinpoint voice accuracy capability country-wide in 243 states, county, and local correctional agencies,” notes Securus in the Pinal County contract.

The enrollment of incarcerated people’s voice prints allows corrections authorities to biometrically identify all prisoners’ voices on prison calls, and find past prison calls in which the same voice prints are detected. Such systems can also automatically flag “suspicious” calls, enabling investigators to review discrepancies between the incarcerated person’s ID for the call and the voice print detected. Securus did not respond to a request for comment on how it defined “suspicious.” The company’s Investigator Pro also provides a voice probability score, rating the likelihood that an incarcerated person’s voice was heard on a call.

Michael Lynch, an intelligence coordinator for the Alachua County Jail in northern Florida, confirmed that his county recently agreed to purchase Securus’s voice recognition program. Lynch said that the voice prints produced by the program will be permanently archived at Securus’s facility in Texas. He said the jail hopes the technology will address the problem of incarcerated people using each others’ personal identification numbers, or PINs. “The problem is inmates that are committing other criminal acts or contacting victims or witnesses and using other inmates’ PIN to do that,” he said in a phone call. “Voice [biometrics] will tell us who’s making the calls.”

Image of Securus’s Investigator Pro VoiceSearch tool from a 2017 approved contract proposal to Pinal County, Ariz.

Image of Securus’s Investigator Pro VoiceSearch tool from a 2017 approved contract proposal to Pinal County, Ariz.

Securus’s voice recognition program can also identify the voices of people outside prisons, both former prisoners and those who have never been incarcerated but communicate with people inside.

New York and Texas state corrections officials confirmed that their agencies retain the voice prints of formerly incarcerated people, like Dukes, allowing them to identify them by name if currently incarcerated people call them in the future.

And New York and Pinal County, Arizona, confirmed that their voice recognition programs can identify the voices of outside callers.

New York’s contract proposal with Securus states that outsiders’ voice samples can be used to “search for all other calls” in their recorded call database to find where those voices occur. In an email, New York prison officials confirmed that this program will give investigators the ability to extract a voice print from an outside caller and use it to “identify that a call recipient has participated in multiple phone calls.” They added that the program will not have names associated with outsiders’ voice prints.

In a statement, Pinal County Sheriff’s Office spokesperson Navideh Forghani also confirmed this outsider voice-tracking capability, noting that while their software does not identify non-incarcerated people by name, it can track “suspicious activities,” such as “multiple inmates speaking to one person on the outside on a reoccurring basis.”

With this technology, a press release for Investigator Pro notes, an investigator can now answer questions like, “What other inmates are talking to this particular called party?” and “Are any of my current inmates talking to this released inmate?”

Prisoners’ rights advocates worry that outsider voice surveillance technology could also be used to coordinate crackdowns against prison organizing campaigns.

“Using this technology to trace the voices of outside callers and flag those who speak with more than one person in a system, staff can use calls with outside organizers to quickly identify the incarcerated activist they support,” said Bianca Tylek, director of the Corrections Accountability Project, which works to curb the influence of commercial interests in the criminal justice system. Tylek noted that during the 2018 national prison strike, corrections staff routinely retaliated against incarcerated activists by using tactics like solitary confinement, job termination, and facility reassignment.


The Pressure to Participate

Advocates assert that corrections agencies have been building up large-scale voice-print databases with limited input from the public or from incarcerated people and their families. While some state corrections agencies have put out public notices to families about payment options for new phone systems, they seldom mention the voice-print databases, which are rarely discussed outside of industry conferences and internal talks with contractors.

“Every time there’s a new contract, there’s new surveillance, but they don’t say anything,” said Tylek. “I’ve never seen authorities post a public notice about new surveillance updates or tell families.”

Keeping their plans opaque has allowed authorities to quietly pressure incarcerated people into giving up their biometric data — or to enroll them without their knowledge. According to Securus’s 2019 Investigator Pro contract with Alachua County, Florida (which includes Gainesville), “Inmates will participate in a covert voice print enrollment process.”

In Texas, state prisoners must enroll in the voice recognition program if they want to make calls. According to Jeremy Desel, a spokesperson for the Texas Department of Criminal Justice, Investigator Pro’s voice enrollment process is “the lock and key” to the Texas state prison phone system. Likewise, in Pinal County, Arizona, phone access is severely limited for prisoners who decline to enroll in the voice recognition program. “If inmates choose not to participate, they can still utilize the phone system but only to make phone calls to their attorneys,” said Forghani, the county sheriff’s office spokesperson.

In some cases, prisoners participate without even knowing, said Martin Garcia, a 33-year-old who is incarcerated at Sing Sing in New York.

“A lot of guys don’t know technology,” he said. “They’ve been in there so long, they’ve never heard of Google.” The voice enrollment procedure, he continued, is seen as “just another thing they follow to talk to their family.”

Garcia was upset to hear that Securus’s voice-tracking capabilities, as described in its approved contract with the New York State Department of Corrections and Community Supervision, could mine prison call databases to identify which other prisoners outside callers had contacted. “Are they criminals just because they’re talking to someone incarcerated?” he said. “To me, you’re criminalizing relationships. Some people may be hesitant to interact with me if they could be put in a database.”

After being briefed by The Appeal and The Intercept about the program, New York State Assembly Member David Weprin publicly called on the state Department of Corrections to give incarcerated people more choice regarding the voice recognition program. At a Tuesday hearing, Weprin, chair of the Assembly’s Committee on Correction, asked the Department of Corrections’ acting commissioner, Anthony J. Annucci, to add a provision that allows incarcerated people with legitimate concerns about voice surveillance to “not be denied phone privileges.” Annucci did not immediately agree to the request, instead pointing out that people have the option to make unmonitored calls to their attorneys.

In a statement to The Appeal and The Intercept, Weprin said he is “concerned with the deployment and use of voice recognition software” in New York state prisons and will be working with his colleagues to further investigate the technology.

Building the Databases

The rapid, secretive growth of voice-print databases is “probably not a legal issue, not because it shouldn’t be, but because it’s something laws haven’t entertained yet,” noted Clare Garvie, a senior associate at Georgetown Law’s Center on Privacy and Technology. “It’s not surprising that we’re seeing this around prisons, just because it can be collected easily,” she continued, referring to biometric voice data. “We’re building these databases from the ground up.”

The scale of prisons’ emerging voice biometric databases has not been comprehensively documented nationwide, but, at minimum, they already hold more than 200,000 incarcerated people’s voice prints.

New York’s Department of Corrections, which incarcerates just under50,000 people, confirmed that approximately 92 percent of its population had been enrolled in the voice recognition system. State corrections authorities for Florida, Texas, and Arkansas, which hold about 260,000 prisoners combined, also confirmed that they are using Investigator Pro’s voice recognition technology. Connecticut and Georgia’s state corrections systems, which incarcerate roughly 13,000 and roughly 52,000 people, respectively, have also purchased Securus’s voice-print technology.

The databases of recorded calls from which prison authorities could search for outsiders’ voice samples could also potentially include millions of recorded calls for state and countywide systems. According to the design requirements New York’s Department of Corrections gave to Securus, for example, the company must be able to record every call, archive all call recordings for a year, and maintain any calls flagged for investigative purposes “indefinitely” through the life of the contract, which ends in 2021. (In the documents, Securus estimated that 7 percent of prison calls made per year would total 1.5 million calls, suggesting that the call database could retain over 20 million calls.)

Greco of the Legal Aid Society says he understands the value of such monitoring capabilities, pointing out that incarcerated people do sometimes have to deal with other prisoners taking their PINs or threatening their families for money. But the extension of this technology into the monitoring of people outside prisons, and the lack of transparency and regulation of these new databases concerns him. If voice prints were shared with police, for example, they could try to compare them with voices caught on a wiretap, he notes, despite scientists’ skepticism about the reliability of voice print matches for criminal prosecutions. New York State’s Department of Corrections declined to answer questions regarding whether it would share the data with other agencies.

Either way, Greco said, there’s cause for concern. “Once the data exists, and it becomes an accepted part of what’s happening, it’s very hard to protect it or limit its use in the future,” he said.

That has implications far beyond prisons, argues Garcia, the man incarcerated at Sing Sing. “First you use this on the people marginalized in society, criminalizing the families of those incarcerated,” he said. “But, especially in Trump’s America, the sky is the limit with this.”

New technology was key to arrest of fugitive for 35 years

Indianapolis - State investigators say the very things that helped a convicted Indiana murderer stay under the radar for more than three decades were what finally got her caught.


When Tennessee police officers saw the 1970s prison photo, they knew Linda Delaney, a well liked, small town grandmother, was in fact Linda Darby, an escapee from the Indiana Women's Prison. She was a wife convicted of settling a family financial hardship by shooting her husband and setting him and their Hammond home on fire.

The interviewer from Eyewitness News' sister station in Nashville asked her if she felt she had been living a lie all these years.

"Uh huh. But when do you stop? Where do you go back to?," said Darby.

Darby will soon be headed back to an Indiana prison to finish her life sentence. The key to her eventual capture, like so many long unsolved crimes, new technology.

"We had multiple analysts who worked on this case," said Monte McKee, Indiana Intelligence Fusion Center.

McKee runs the Fusion Center under the Indiana Department of Homeland Security. It was just ten days ago its state and federal analysts re-entered the data from the cold 1972 escape case of Linda Darby. 

In less than the equivalent of three workdays they were focused on a retirement aged Tennessee wife and mother of two. A woman who shared the first name of Linda, whose registered birth-date and social security number were just a few digits off their suspect's. Those were changes that might be easy for a fugitive to remember and avoid attracting attention.

McKee adds, "It was her undoing in this case as well."

Darby never got another driver's license, another red flag used to pick her from a database of millions of people. Officials hope to follow up this closed case with others. The Department of Corrections brought the Fusion Center analysts another 300 cases of missing prisoners and parolees.

China to build giant facial recognition database to identify any citizen within seconds

Project aims to achieve an accuracy rate of 90 per cent but faces formidable technological hurdles and concerns about security

Facial recognition system for visitors is being used at China’s border control. 

Facial recognition system for visitors is being used at China’s border control. 

China is building the world’s most powerful facial recognition system with the power to identify any one of its 1.3 billion citizens within three seconds.

The goal is for the system to able to match someone’s face to their ID photo with about 90 per cent accuracy.

The project, launched by the Ministry of Public Security in 2015, is under development in conjunction with a security company based in Shanghai.

The system can be connected to surveillance camera networks and will use cloud facilities to connect with data storage and processing centres distributed across the country, according to people familiar with the project.

However, some researchers said it was unclear when the system would be completed, as the development was encountering many difficulties due to the technical limits of facial recognition technology and the large population base.

At present, similar systems operate on a smaller level, including police databases and city or provincial ID pools.

But these operate separately and are on a much smaller scale.There is also a national database of police suspects and people of interest to the government.These may continue to be used independently after the national system is established.


The core data set for the national system, containing the portrait information of each Chinese citizen, amounts to 13 terabytes.

The size of the full database with detailed personal information does not exceed 90 terabytes, according to technical documents on the ministry’s website and a paper written by police researchers.

Chen Jiansheng, an associate professor at the department of electrical engineering at Tsinghua University and a member of the ministry’s Committee of Standardisation overseeing technical developments in police forces, said the system would have to be built on an unprecedented scale because no country had a population as big as China’s.

The system was being developed for security and government uses such as tracking wanted suspects and public administration, he said.

Commercial application using information sourced from the database will not be allowed under current regulations.

“[But] a policy can change due to the development of the economy and increasing demand from society,” Chen said.

Giving commercial sectors access to the database under proper regulation would create new business opportunities by helping to improve customer service, he said.

Chinese companies are already taking the commercial application of facial recognition technology to new heights.

With a smile or blink of the eyes to a camera, students can now enter their university halls, travellers can board planes without using a boarding pass and diners can pay for a meal at KFC.


Some other restaurants have even offered discounts to customers based on a machine that ranks their looks according to an algorithm. Customers with “beautiful” characteristics – such as symmetrical features – get better scores than those with noses that are “too big” or “too small” and those that get better scores will get cheaper meals.

Some public lavatories in Beijing also use facial recognition so that the automatic dispensing machines will deny toilet paper to people who ask for it more than once within a given period.

Facial recognition could supersede other personal identification methods that are used to make payments such as scanning fingerprints or QR codes on a mobile phone.

But the government project has prompted controversy among artificial intelligence experts.

Cheng Mingming, a professor of computer science at Nankai University in Tianjin, said that despite the scale of the project, technological advances meant that all the information could be stored in small, portable drives – which raised the risk of data theft.

He said a palm-sized commercial hard drive nowadays could store 10 terabytes or more of data and you could “pack it in a suitcase and board a flight”.“If the facial data and related personal information is stolen and put on the internet, it will cause big problems,” Cheng said.

For instance, due to the rapid advance of facial recognition technology, a person or organisation could take a photo and identify strangers at a party or on the street without their knowledge, Cheng said.

But a network security vendor for the Ministry of Public Security dismissed the possibility.“To download the whole data set is as difficult as launching a missile with a nuclear warhead. It requires several high-ranking officials to insert and turn their keys at the same time,” the vendor said.


The 1.3 billion-person facial recognition system is being developed by Isvision, a security company based in Shanghai.

Isvision confirmed to the South China Morning Post that it had won the contract last year but declined to provide details.

“The progress of development is confidential. At present we have no information for public disclosure,” a company spokesperson said.

Isvision security cameras with facial recognition capabilities were first deployed in Tiananmen Square as early as 2003, according to the company’s website.

The system was connected to the police database of suspects, capable of recognising and tracking potential targets in a large crowd.

The company has also set up similar systems for law enforcement authorities in Xinjiang and Tibet, where riots have broken out from time to time because of serious ethnic conflicts.

According to Fan Ying, a researcher at the ministry’s population management research centre in Beijing, the project team has encountered “unprecedented challenges” due to the government’s high demands for speed and accuracy.

When a photo, gender and age range are inputted, the system is required to find a match within three seconds with an accuracy level higher than 88 per cent.

Fan and colleagues tested the facial recognition algorithm developed by Tsinghua University, a world-leading institute in this field of research, and they were disappointed with the results.

They found that the accuracy of the photo that most closely matched the face being searched for was below 60 per cent. With the top 20 matches the accuracy rate remained below 70 per cent, Fan and collaborators reported in a paper published in the domestic journal Electronic Science and Technology in May.

“It cannot solve problems with real-life applications,” they added.

The system developed by Isvision will use an algorithm developed by SeetaTech, a start-up established by several researchers from the Institute of Computing Technology at the Chinese Academy of Sciences in Beijing.

SeetaTech confirmed to the SCMP its involvement in the national facial recognition project but declined to comment further.

A researcher at the Institute of Computing Technology familiar with the project said some huge technical hurdles remained.

“Among 1.3 billion people, some totally unrelated people have faces so alike even their parents cannot tell them apart,” the researcher said.

“Currently the access to the database is limited to a few security companies with very close ties with the Ministry of Public Security.“More access will definitely lead to higher risk of [data] leakage.”

The researcher warned that the cost of the convenience facial recognition could bring to everyday life was “sacrificing security”.