The statement comes after FedScoop reported Thursday that the DOJ disclosed in its public inventory of AI use cases that the FBI was initiating use of Rekognition as part of something called “Project Tyr.” The disclosure is significant because Amazon had previously extended a moratorium on police use of Rekognition, though the company did not originally clarify how that moratorium might apply to federal law enforcement. 

In an emailed response to FedScoop, Amazon spokesperson Duncan Neasham said: “We imposed a moratorium on police departments’ use of Amazon Rekognition’s face comparison feature in connection with criminal investigations in June 2020, and to suggest we have relaxed this moratorium is false. Rekognition is an image and video analysis service that has many non-facial analysis and comparison features. Nothing in the Department of Justice’s disclosure indicates the FBI is violating the moratorium in any way.”

According to Amazon’s terms of service, the company placed a moratorium on the “use of Amazon Rekognition’s face comparison feature by police departments in connection with criminal investigations. This moratorium does not apply to use of Amazon Rekognition’s face comparison feature to help identify or locate missing persons.”

The company’s public statement about its one-year moratorium in 2020, which was reportedly extended indefinitely, stated that it applied to “police use of Rekognition.” That statement did not specifically call out the “face comparison feature” or use of the tool related to criminal investigations.

Neasham further stated on Friday that Amazon believes “governments should put in place regulations to govern the ethical use of facial recognition technology, and we are ready to help them design appropriate rules, if requested.”

The description of the use case in DOJ’s AI inventory doesn’t mention the term “facial recognition,” but it states that the agency is working on customizing the tool to “review and identify items containing nudity, weapons, explosives, and other identifying information.” Neither Amazon nor the DOJ have clarified FedScoop questions about whether the FBI had access to facial recognition technology through this work.

Civil liberties advocates told FedScoop that the use case surprised them, given Amazon’s previous statements on facial recognition, Rekognition, and police.

“After immense public pressure, Amazon committed to not providing a face recognition product to law enforcement, and so any provision of Rekognition to DOJ would raise serious questions about whether Amazon has broken that promise and engaged in deception,” American Civil Liberties Union of Northern California attorney Matt Cagle said in a Thursday statement to FedScoop.  

DOJ spokesperson Wyn Hornbuckle did not address several aspects of the project but provided a statement pointing to the agency’s creation of an Emerging Technologies Board to “coordinate and govern AI and other emerging technology issues across the Department.” The FBI declined to comment through the DOJ.

The post Amazon says DOJ disclosure doesn’t indicate violation of facial recognition moratorium appeared first on FedScoop.

The caucus, comprised of Sens. Martin Heinrich, D-N.M., Rob Portman, R-Ohio, Brian Schatz, D-Hawaii, Cory Gardner, R-Colo., Gary Peters, D-Mich., and Joni Ernst, R-Iowa, will help members of Congress and their staff connect with outside artificial intelligence experts.

“AI is likely to be one of the most transformative technologies of all time,” Portman said in a statement. “AI is a mix of promise and pitfall, which means, as policymakers, we need to clear-eyed about its potential.” The caucus will work to make sure that “Congress is home to the substantive conversations necessary to make responsible policy about emerging technology and ensure AI works for, and not against, American citizens and U.S. competitiveness,” he added.

The formation of the caucus comes about a month after President Donald Trump signed an executive order outlining his administration’s AI national plan. The “American AI Initiative” lays out a “multi-pronged approach” to the challenge of continued American leadership in this technology. It directs agencies to prioritize AI in their research and development budgets, make data available to researchers, create training programs to help workers gain AI-relevant tech skills and more.

Congress has faced various questions about the implications of AI-powered technologies over the past year. For example, in July, the American Civil Liberties Union published a study that put sitting members of Congress in the crosshairs of misidentification by facial recognition technology. Using Amazon’s Rekognition software, the group compared a photo set of sitting members of Congress to a database of publicly available arrest photos. The software, according to ACLU, incorrectly identified 28 sitting members of Congress as individuals who have been arrested. Lawmakers, predictably, had some questions.

Amazon argued that the study is actually misleading, and that ACLU used incorrect settings to get their results. “Machine learning is a very valuable tool to help law enforcement agencies, and while being concerned it’s applied correctly, we should not throw away the oven because the temperature could be set wrong and burn the pizza,” Matt Wood, AWS’ head of machine learning, wrote in a blog post. “It is a very reasonable idea, however, for the government to weigh in and specify what temperature (or confidence levels) it wants law enforcement agencies to meet to assist in their public safety work.”

The task of weighing in remains ahead — the senators indeed say they will consider how AI developers can maintain “important ethical standards.”

The post Senate’s new AI caucus will work toward ‘responsible policy’ appeared first on FedScoop.

The study raises still more questions about race- and gender-based bias in the algorithms that Amazon has been continuously attempting to sell to federal agencies and police departments.

Facial analysis is a service that Amazon Rekognition offers, but it’s not the same as facial recognition. Per Rekognition’s FAQ page, “facial analysis is the process of detecting a face within an image and extracting relevant face attributes from it.” The software can, for example, tell the user the gender and emotions of the face, as well as things like whether it is wearing glasses or has facial hair.

Facial recognition, by contrast, “is the process of identifying or verifying a person’s identity by searching for their face in a collection of faces.”

This distinction is Amazon’s first of several pushbacks on the results of the study. “It’s not possible to draw a conclusion on the accuracy of facial recognition for any use case — including law enforcement — based on results obtained using facial analysis,” Matt Wood, the general manager of artificial intelligence at Amazon Web Services, said in a statement emailed to FedScoop. He added that the study used an outdated version of Rekognition, and the algorithms have since been updated and improved.

Wood said that a similar internal test of the facial recognition component yielded very different results. “Using an up-to-date version of Amazon Rekognition with similar data downloaded from parliamentary websites and the Megaface data set of 1 million images, we found exactly zero false positive matches with the recommended 99 percent confidence threshold,” he said.

But this distinction is unlikely to assuage the concerns of civil rights and civil liberties advocates, who worry that there isn’t enough regulation in place to protect against a kind of compounding human and algorithmic bias.

“Even if the Amazon Rekognition services and products you are selling to police departments were completely flawless, the potential for abuse on historically marginalized communities would not be reduced,” MIT Media Lab researcher Joy Buolamwini wrote in a letter to Amazon in June.

Interestingly, the MIT study isn’t about bias in facial analysis algorithms per se, but about how companies respond to the publication of studies that show biased results. It shows how Microsoft, Face++ and IBM, all purveyors of software that Buolamwini tested in a previous study, responded to that study by releasing improved algorithms. In this context, Amazon Recognition, and Miami-based company Kairos, represent a kind of control group.

The study finds that publicly discussing bias seems to be an effective mechanism to get companies to reduce that bias. “By highlighting the issue of classification performance disparities and amplifying public awareness, the study was able to motivate companies to prioritize the issue and yield significant improvements within 7 months,” it states.

The study also represents a growing interest in auditing facial recognition and artificial intelligence algorithms. The National Institute of Standards and Technology recently released the results of such an audit of selected facial recognition algorithms and found that the algorithms are getting better over time.

Wood expressed Amazon’s support for this line of study: “We continue to seek input and feedback to constantly improve this technology, and support the creation of third-party evaluations, data sets, and benchmarks,” he said. “Accuracy, bias, and the appropriate use of confidence levels are areas of focus for AWS, and we’re grateful to customers and academics who contribute to improving these technologies.”

The post Study finds biases in Amazon Rekognition’s facial analysis tool appeared first on FedScoop.

In the resolution, some shareholders ask the company’s board of directors to step in and stop such sales “unless the Board concludes, after an evaluation using independent evidence, that the technology does not cause or contribute to actual or potential violations of civil and human rights.” The petitioning shareholders expect the resolution will be voted on during the company’s annual meeting this spring.

The primary shareholder on the petition is the Sisters of St. Joseph of Brentwood, a member of the Tri-State Coalition for Responsible Investment.

“The Sisters of St. Joseph of Brentwood have committed as a congregation to support immigrant communities and promote racial equity, and this applies to our investments,” Sister Patricia Mahoney said in a statement. “Therefore, we filed this proposal because we are concerned that Amazon has pitched facial recognition technology to Immigration and Customs Enforcement (ICE) and piloted its Rekognition with police departments, without fully assessing potential human rights impacts.”

The group isn’t alone in their interest in halting, or at least slowing, government use of these technologies.

Just earlier this week, American Civil Liberties Union, together with a bunch of other civil rights groups, called on Amazon and other software companies to stop selling facial recognition tech to the government. The ACLU seems pleased to see the issue raised by yet another set of stakeholders.

“The fact that Amazon’s shareholders felt compelled to take this up to the company’s board of directors should be a wake-up call to Amazon’s leadership to take concerns around face surveillance seriously,” Shankar Narayan, the Technology and Liberty Project director of the American Civil Liberties Union of Washington, said in a statement.

Despite regular pressure from civil rights groups, and increasing questioning from Congress, Amazon has maintained that facial recognition technology is a net positive for the government. Asked if the business has “drawn any red lines” around its government work after employees protested the sale of Rekognition to law enforcement agencies, Teresa Carlson, vice president of worldwide public sector for Amazon Web Services, said no.

“We provide them the tools, we don’t provide the solution application that they build,” she said of Amazon’s government sales during the Aspen Security Forum in July 2018. “And we often don’t know everything they’re actually utilizing the tool for. But they need to have the most innovative and cutting-edge tools they can.”

It’s unclear which additional shareholders are joining the Sisters of St. Joseph of Brentwood in their resolution, but the organization says that the group represents a total of $1.32 billion in assets under management.

The post Shareholders also want Amazon to stop selling facial recognition tech to government appeared first on FedScoop.

The group, which includes Sen. Ed Markey, D-Mass., as well as Reps. Ro Khanna, D-Calif., John Lewis, D-Ga. and more, is asking for responses to four broad questions about Rekognition’s makeup and use. Specifically, they want to know how Amazon tests the technology for accuracy and bias, what protections are built in, whether Amazon conducts audits of how customers — especially law enforcement agencies — use the tech and which government customers are using the system.

“Facial recognition technology may one day serve as a useful tool for law enforcement officials working to protect the American public and keep us safe,” the lawmakers write. “However, at this time, we have serious concerns that this type of product has significant accuracy issues, places disproportionate burdens on communities of color, and could stifle Americans’ willingness to exercise their First Amendment rights in public.”

This isn’t the first time these lawmakers have expressed concern. In July, after the American Civil Liberties Union conducted a study using Rekognition in which in 28 sitting members of Congress were falsely identified as individuals who have been arrested for a crime, Markey and Rep. Jimmy Gomez, D-Calif., each sent a letter on the subject.

Markey’s original letter contained seven questions for Amazon to answer — per the most recent letter it seems Amazon failed to provide “sufficient” answers to three of these.

Despite growing concern about the use and potential for abuse of facial recognition technology, Amazon has maintained that tools like Rekognition have a net benefit for government. Asked if the business has “drawn any red lines” around its government work after employees protested the sale of Rekognition to law enforcement agencies, Teresa Carlson, vice president of worldwide public sector for Amazon Web Services, said no.

“We provide them the tools, we don’t provide the solution application that they build,” she said at the Aspen Security Forum in July. “And we often don’t know everything they’re actually utilizing the tool for. But they need to have the most innovative and cutting-edge tools they can.”

The lawmakers are looking to hear back from Bezos by Dec. 13.

The post Lawmakers demand answers on Amazon Rekognition (again) appeared first on FedScoop.