Читать книгу The Cost of Free Shipping - Группа авторов - Страница 27

Amazon’s 2018 annual report does not state any concerns over customer or user “privacy.” In fact, the only part of the report to address privacy pertains to how Amazon’s endangering of individual users’ privacy poses a risk to the company through increased governmental regulation and potential lawsuits and criminal penalties.³⁴ In other words, violations of privacy threaten Amazon’s bottom-line first. This indifference to customer privacy is telling. Amazon’s many, widely used products (e.g., Echo, Rekognition, and Ring) place the corporation at the center of intense debates about individuals’ privacy. Amazon stands accused of both violating the privacy of its users and customers (including minors), of facilitating such violations, and expressing only mild concern over such violations.³⁵

For example, Amazon sells a popular line of speakers called Echo. The many Echo models have a special feature: a voice-activated personal assistant. This assistant is personified as “Alexa” and answers any questions and requests that are made of it. If someone audibly utters the Echo’s “wake word”—by default “Alexa”—then it will respond to commands verbalized afterward. In order to do this, Echo is “always-on,” awaiting any activation request. Echo also records audio after the wake word is uttered and gathers information on the user’s location, sending this information to Amazon for review. According to Amazon, “Alexa should remember context and past interactions,”³⁶ to help its artificial intelligence (AI) to self-improve. Additionally, Amazon employees manually review these recordings for the purpose of improving Alexa, too.

The Echo introduces a variety of privacy concerns that neither Amazon nor society at large has been able to successfully answer. Users do not choose whether Amazon records their instructions to Alexa—as this recording is automated, there may be people present in a room with an Echo device that are unaware they are being recorded by Echo. Consequently, Amazon controls an incredible cache of audio, constituting a potentially huge database of customer desires ripe for economic exploitation by the highest bidder. Problematically, users have discovered a surprising number of false-positive activations from non-wake words. In other words, Alexa may activate itself without actually being summoned by a user and then proceeded to record conversations and other speech that was never intended to be recorded. In the summer of 2019, it was revealed that Apple’s AI “Siri” had recorded people having sex and engaging in drug deals, all without the individuals’ knowledge; similar things have occurred with Echo. Echo’s AI can identify a user’s mood or emotions from their voice, and thus respond in kind. Since the Alexa assistant helps users to connect to other smart devices (which involve additional purchases), interactions with these devices are also recorded. Typically, a purchase request made to Alexa results in the assistant purchasing a product from Amazon.com, thus seamlessly integrating Amazon into consumption patterns. Thus, Echo facilitates a corporate marketer’s dream—having direct access to the unconscious and often only vaguely articulated desires of customers. This requires massive surveillance. When individuals share aspects of their personal lives—characteristics, preferences, interests, and aspirations—such data is easily commodified and sold to marketers and other corporations seeking to sell individuals additional products and services. This is what communication scholar Emily West refers to as “surveillance as a service,” and the Electronic Privacy Information Center argued before the Federal Trade Commission that Echo constituted “unlawful surveillance under federal wiretap law.”³⁷

Amazon has developed a facial recognition platform called Rekognition. Unsurprisingly, some of the most interested and enthusiastic customers are governments. The potency of technology able to compare video-recorded individuals to information stored in state-based databases opens up a panorama of potential abuses. In addition to being able to recognize and identify individual’s faces, Rekognition can also identify clothing and discern gender. The platform can track many people (perhaps hundreds consecutively) through crowds, identify what individuals are doing, discern their emotional state (e.g., happy, sad, or fearful), identify non-human objects, and read words (e.g., on signs, license plates). Most concerning of all, is that Rekognition also can be used to flag “unsafe” or “inappropriate” things. Amazon has marketed Rekognition to local U.S. police forces; since AWS already hosts the body-camera and surveillance camera footage from many police departments, Rekognition can be an “add-on” feature for a mere $6–12 a month extra. Such robust and integrated facial recognition platforms are obviously able to assist authoritarian states, but “democratic” states are equally able to exploit Rekognition for their own unchallengeable advantage. Many obvious concerns have been articulated by civil liberties and privacy organizations like the American Civil Liberties Union and the Electronic Frontier Foundation. These advocates fear that Rekognition raises the risks for already over-policed populations (especially the poor and people of color), can invasively track and manipulate immigrants, and be used to identify and arrest protesters and activists.

Amazon deflects such criticisms by instead pointing to Rekognition’s allegedly “positive” uses, such as finding lost children. The selective highlighting of an invasive technology’s “social good” is an old, established strategy, just as states seek to negatively frame privacy-protecting technologies that limit state power. For example, public-key encryption protocols have been regularly critiqued by governments for fears that it could empower pedophiles, drug-dealers, or terrorists. Instead, governments advocate flawed encryption platforms for which they hold “backdoor keys,” that they proclaim will only be used when absolutely necessary (e.g., with a court-signed warrant). The problem with these arguments is that backdoored encryption is known to be flawed and will be avoided by those seeking to avoid state surveillance. Robust, well-functioning encryption empowers people against state power. But, Rekognition and other facial recognition tools give a clear political advantage to states, and suppress free expression and existence in public.

Ring is a technology that integrates a doorbell with a microphone and video camera, allowing homeowners to view their front doors and be alerted—even from afar—when someone rings the doorbell. It can be activated upon ringing and detect motion; upon activation, the camera footage is recorded to Amazon’s cloud, as well as sent to individuals’ smartphones. Problematically, Ring can stoke or enhance homeowners’ paranoia about safety and security, thus provoking the purchase of more Amazon cameras. Many people pass by or approach front doors on a daily basis—to deliver mail or packages, visit residents, drop off fliers, or ask for directions. Additionally, people soliciting donations or selling products come to front doors, often (but not always) ringing doorbells. Most homeowners are unaware of how often this occurs and Ring can generate suspicions that all the above individuals could be potential criminals seeking to break in, assault, or rob those in the house. Owners can use Ring’s “neighborhood watch” app to post messages about “suspicious” people they witness on their Ring cameras. Disproportionately, Ring owners post such messages about people of color, upon whom racist stereotypes are focused. As Ring CEO Jamie Siminoff put their mission: to “declare war” on “dirtbag criminals.”³⁸

Because Ring generates fears related to crime, police are drawn to this technology, too. Amazon has partnered with over 400 local law enforcement agencies in the U.S.³⁹ Police can use their Amazon-provided portals to see maps of neighborhood Ring users and request video footage from homeowners. Some of these Ring cameras also capture activity taking place in public space, not just an owner’s private property. Consequently, the Electronic Frontier Foundation calls Ring a “perfect storm of privacy threats.”⁴⁰ It helps to create a wider surveillance network for police, controlled by paranoid (and biased) homeowners, with infrastructure owned by Amazon. If homeowners cannot be convinced by police—who are trained to strategically apply such pressure—to hand over their Ring’s camera footage, police can simply subpoena Amazon directly for footage.

A hegemonic threat to privacy may emerge if all the “smart devices” consumers have placed in their homes can become networked and then integrated into state surveillance systems. Unfortunately, something like this may be on the horizon, with Amazon, Google, and Apple beginning to collaborate on connecting their respective products together, under the name “Project Connected Home over IP.” At present, it is unclear whether this will make it easier or harder for unpermitted access to users’ systems, but it absolutely invites further corporate intervention in users’ homes—and all the considerable risks that involves.⁴¹