|
Source For information about the software and how it works, click here: here.
FBI under fire over secret project to recognise people by their TATTOOS By Stacy Liberatore For Dailymail.com Published: 14:39 EST, 3 June 2016 | Updated: 18:29 EST, 7 June 2016 Privacy campaigners have revealed new details of a government-funded program that they say ‘raises significant First Amendment questions’. The Electronic Frontier Foundation found that the FBI is developing software that looks at tattoos to determine people's gang affiliation, sub-cultures, religious beliefs or political ideologies. The technology deciphers the meaning of tattoos, maps connections between people with similar themed tattoos and makes inferences about people from their ink. The FBI is developing software that will also look at body art, but to determine gang affiliation, sub-cultures, religious beliefs or political ideologies. These algorithms are designed to decipher the meaning of tattoos and also map connections between people with similarly themed tattoos or make inferences from their tattoos The FBI is developing software that will also look at body art, but to determine gang affiliation, sub-cultures, religious beliefs or political ideologies. These algorithms are designed to decipher the meaning of tattoos and also map connections between people with similarly themed tattoos or make inferences from their tattoos WHAT CAN THE ALGORITHMS DO? 1. Tattoo Similarity: Match tattoos from different people that share visual elements or symbolism 2. Mixed Media: Match tattoos to a similar image in another medium 3. Region of Interest: Match a part of a tattoo to the full tattooed area 4. Tattoo Identification: Match images of the same person’s tattoo over time 5. Tattoo detection: Determine whether an image contains a tattoo or not During the investigation, the team learned that the National Institute for Standards and Technology (NIST) teamed up with the FBI in 2014 to 'promote and refine automated tattoo recognition technology' -- which EFF says violates privacy,free speech and the right to associate. Electronic Frontier Foundation (EFF), a civil rights advocacy group, has been filing public records requests to show how law enforcement agencies are using biometric technology to profile individuals. 'The NIST project is about measuring the effectiveness of algorithms for accurately matching digital images, not about religious affiliations,' a spokesperson from the National Institute for Standards and Technology told DailyMail.com in an email. 'The NIST project is also not about the many complex law enforcement policies or approaches that may be related to images of tattoos.' 'We are reviewing the EFF report and will carefully consider their concerns.' WHY IS THIS SOFTWARE DEEMED UNETHICAL? The algorithms are designed to not only decipher the meaning of tattoos, but also maps connections between people with similarly themed tattoos or make inferences about people from their tattoos. The agency obtained 15,000 images of tattoos from arrestees, during the experiment, which were handed out to third parties without inmates consent. NIST is said to have shared the tattoo base with a total of 19 other organizations: five research institutions, six universities and eight private companies -- including MorphoTrak, one of the largest marketers of biometric technology to law enforcement agencies. During the initial experiments, participants were required to perform a series of rest sand share their results with NIST. These tests included whether the algorithm could identify a tattoo in an image and if it was able to match different images of the same tattoo taken over time. However, what EFF found most alarming was that the research involved matching common visual elements between tattoos with operational goal of establishing connections between individuals. Tattoos are considered biometric characterizes, but are unique because they were not given at birth and are a form of expression. However, what is more important about this body art in EFF's investigation is that tattoos are also seen as a form of speech. 'Any attempt to identify, profile, sort, or link people based on their ink raises significant First Amendment questions,' reads the report from EFF. 'The research program is so fraught with problems that EFF believes the only solution is for the government to suspend the project immediately.' Beginning this project, which is called Tatt-C, the agency obtained 15,000 images of tattoos from arrestees, which were handed out to third parties without inmates consent, according to the EFF report. NIST is said to have shared the tattoo database with a total of 19 other organizations: five research institutions, six universities and eight private companies to test the algorithm-- including MorphoTrak, one of the largest marketers of biometric technology to law enforcement agencies. NIST had no role in the collection of the dataset, which consists of data obtained by the FBI,' said MIST. 'Before beginning the project, NIST confirmed that the FBI had the authority to collect the images and provide them to NIST.' 'The agreement specified that NIST should receive no private information on the individuals whose tattoos were imaged'. 'The tattoo dataset was not distributed by NIST. ' 'NIST takes privacy concerns seriously and after we were alerted to the concerns by EFF regarding one of the images in the dataset, NIST conducted a review and has found several other images that raise similar concerns. NIST has alerted the FBI about this.' Electronic Frontier Foundation (EFF), a civil rights advocacy group, has been filing public records requests to show how law enforcement agencies are using biometric technology to profile individuals. They found the NIST teamed up with the FBI in 2014 to 'promote and refine automated tattoo recognition technology' Electronic Frontier Foundation (EFF), a civil rights advocacy group, has been filing public records requests to show how law enforcement agencies are using biometric technology to profile individuals. They found the NIST teamed up with the FBI in 2014 to 'promote and refine automated tattoo recognition technology' During the initial experiments, participants were required to perform a series of tests and share their results with NIST. These included whether the algorithm could identify a tattoo in an image and if it was able to match different images of the same tattoo taken over time. 'NIST’s work on this project has been reviewed and determined to not meet the criteria for human subjects research as defined by federal regulations,' said NIST. 'The database contains images only, with no accompanying information on the individuals whose tattoos were photographed.' EFF noted that many of the images they reviewed contained personally identifying information, including names, faces and birth dates. MICROSOFT IS BRINING MINORITY REPORT TO LIFE Microsoft is developing a program that has the potential to mimic the powers of the 'precogs' from the 2002 film Minority Report and stop inmates from becoming recurring offenders. This new technology could predict which inmates will return to jail within six months of their release, based on historical data. 'It's all about software that has the ability to predict the future,' said Jeff King, a senior program manager at Microsoft, in a webcast with police departments earlier this year that has virtually gone unnoticed. The program, which has not been named yet, is still in early development stages and the data used in the proof-of-concept isn't from real inmates. An algorithm will be used analyze specific data of each inmate, such as their crimes or their behavior while incarcerated. King noted this works with up to 91 percent accuracy based on historic data. Microsoft has used predicting technology in the past, but only with predicting a gamers next move. In the Halo video games, the software can predict the users next move, based on the way they have reacted in similar situations in the past. Although this technology is a police officer's dream, some civil rights groups believe this will create categorical suspicion on certain neighborhoods or people who share a set of characteristics. Gary Marx, professor emeritus of sociology at the Massachusetts Institute of Technology, explains this idea as 'tyranny of the algorithm'. 'Every way of looking is also a way of not looking,' said Marx. He believes this type of technology won't push officers to gather accurate human intelligence and just rely on predictions. 'Over time, it could present a 'fallacy of confusing data with knowledge,' Marx argues in his forthcoming book about the shortcomings of data to solve the world's problems. 'If that wasn't alarming enough, NIST researchers also failed to follow protocol for ethical research involving humans—they only sought permission from supervisors after the first major set of experiments were completed,' said the EFF team.. 'These same researchers have also not disclosed to their supervisors that the tattoo datasets they are using to seed the experiments came from prisoners and arrestees.' Another unsettling part of this case is that the NIST and the FBI are not treating these inmates like humans, but instead 'treating inmates as a bottomless pool of free data', EFF said. And the federal research guidelines state that any research involving prisoners must not exploit them, or it is deemed as unethical. 'NIST researchers handling the project initially thought that the review of the FBI’s authority to collect and share the database of images with NIST was the only review required,' explained NIST. 'When it was determined that a human subjects review was necessary, the project was reviewed and determined to not meet the criteria for human subjects research as defined by federal regulations.' +3 During the initial experiments, participants were required to perform a series of tests and share their results with NIST. These included whether the algorithm could identify a tattoo in an image and if it was able to match different images of the same tattoo taken over time However, what EFF found most alarming was that the research involved matching common visual elements between tattoos with operational goal of establishing connections between individuals. Next month, NIST plans to start the second stage of the program called, Tattoo Recognition Technology Evaluation, or Tatt-E. NIST and the FBI will compile as many as 100,000 tattoo images from the Michigan State Police, Tennessee Department of Corrections and Pinellas County Sheriff's Office in Florida, which will be able to test the current recognition systems on a much broader scale. 'Tatt-E is the follow-on to the Tatt-C 2015 activity, which provided researchers with an open dataset of operationally collected tattoo images and use cases to advance image-based tattoo recognition research and development', explains NIST. Psychologists at the University of Twente set out to investigate 'whether perceiving an automated interview system as operated by a human or computer, influences cues to deception in the form of increased sympathetic nervous system activity (SNS)'. During the interrogation, subject's electrodermal activity (EDA) was measured, which is the most commonly used measure in a lie detection test. And EDA reflects SNS activity and is easily measured in one measurement, said researchers. Skin conductance of each subject was also taken in order to see just how much each person was sweating. Each subject was ask 10 questions and then given questionnaires about if they thought the AI was human or computer controlled. Researchers found it was impossible to tell whether or not they were telling the truth or lying if the subject believed the AI was computer controlled. The EDA differences between the truth and a lie and intention condition became larger when participants believed the avatar was controlled by a human. 'Tatt-E seeks to assess comparative and absolute accuracy, along with run-time measures on larger, operationally realistic datasets than seen in Tatt-C.' EFF see two potential uses from this government-funded project. First, the recognition system can uses tattoos to identify suspects over time and eventually do away with fingerprints and traditional facial photos. But the more advanced systems will identify tattoo patterns linked to gang affiliations. The second use that EFF found most unnerving is that it can threaten people's free expression. 'Tattoos are free speech that we wear on our skin,' the group writes. 'Our tattoos express who we were, who we are, and who we hope to be,' states EFF. 'But when law enforcement looks at our tattoos, they see unique biometric identifiers and a shortcut to learning our personal beliefs and our social connections.' 'NIST has indicated that it is looking more closely at the project, but has given no public indication that it will take any action to delay or suspend the program—except for removing questionable presentations from its website.'
|