In a digital age, during a pandemic, virtual classes have exploded in demand. With virtual classes comes virtual assignments, quizzes, and exams, which has led to a concern for academic honesty. In order to make sure that academic honesty is enforced, universities have turned to online proctoring with proctoring software like Respondus’ LockDown Browser, in order to address it. The way that proctoring software operates, however, has made advocacy groups—professors, and students alike—concerned over the well-being of students and the collection of their personal data. Yet others are less concerned about how invasive proctoring software is becoming. California State University, Fullerton’s lecturer Randy Hoffman states, “there isn’t anything to worry about unless a student is trying to beat the system.”1 With students’ personal data or with their health, ProctorU Media Coordinator Franklin Hayes compares the concerns over student privacy to be like “having a cable guy at your house.”2 However, despite these dismissals and the software’s beneficial services, proctoring software brings major risks to student data privacy with the use of artificial intelligence and the data storage it requires, as well as risks to the well-being of students when they take exams.
First, let’s understand what proctoring is. Proctoring is how professors monitor students who take an exam in-class. Professors monitor the testing environment to make sure the exams are taken appropriately by each student. Online Proctoring (or e-proctoring) systems behave like regular proctoring; however, they differ between one another. Honorlock or Proctorio, for example, handle proctoring by having students install an add-on to their browser that controls what they can do in it, while LockDown Browser instead uses a custom browser that takes over the students’ computer screen. Each proctoring system “prohibits students from opening new tabs during [exams]” and has the ability to “[use] the student’s webcam to record a video while they complete the assessment and flag suspicious activity, such as background noise or the student’s eyes shifting away from the computer screen.” Professors or exam proctors can then review these video recordings for possible infractions.3 It sounds like normal proctoring, but just online; however, there are a few issues on how proctoring software works that raises concerns over student data and health.
Proctoring software makes use of computer webcams and microphones to make recordings. By doing so, the software can catch a student looking at a sticky note on their monitor, or capture voices that might provide answers to test questions. Without such software, professors can’t be sure whether the exam has been tampered with or not by a student. While they monitor students, the software uses additional “features” to validate the recordings, such as artificial intelligence (or AI), to determine whether the person taking the exam is really that student, and it will flag any suspicious activity throughout the session automatically. Some advocacy groups have argued that having students turn on their webcams and microphones to a stranger, or even to an algorithm “violate[s] their civil rights.”4 The Electronic Frontier Foundation (or EFF), a non-profit organization that focuses on data privacy, and Mutahar Anas, a YouTuber that does virus investigations and software analyses, both call proctoring systems “the equivalent to spyware.”5 California Community College Board of Governors member Colm Fitzgerald, who works at Delta College in Stockton, CA, told the university’s wire that these systems can be “extremely restrictive.” He related a story about a student who, while taking an e-proctored test, had his brother simply walk in the room, and when he turned around to tell him to leave, he was faulted for violating the testing rules, and failed his exam.6 This opens the door for a student to get their exam revoked by the AI due to someone walking into the room, forgetting that a exam is in session, and the AI thinking the student is cheating due to it.
A point can be made that proctoring companies need this data from their software’s use in order for their AI systems to recognize new behaviors, but this can be resolved by having the student either opt-in or opt-out of data collection with no consequences prior to the exam starting. Other AI programs will ask users if they are OK with the software collecting their data, and if the users says no, they can still continue to use the software with no issues. Proctoring software, on the other hand, makes it mandatory to opt-in, and not agreeing to the terms can lock a student out of their exam. While this data collection makes sense, proctoring companies could add an option that would allow students to send their recordings to the companies, and let them take their exams. That option might limit how many samples the proctoring companies might get for their AI needs, but it should be sufficient to keep improving the algorithm, and might make them data privacy-compliant in the European Union and in the US. EU lawmakers have demanded an investigation to see “whether online proctoring software violates student privacy rights, arguing that it is unfair for universities to force students to use exam monitoring tools that capture their personal biometric data.”11 US senators Richard Blumenthal, Ron Wyden, and Chris Van Hollen wrote to ExamSoft, asking the company “how [student] data is being used before, during, and after tests, by [the company], the virtual proctors, and testing administrators” and how it complies with student privacy laws like the Family Educational Rights and Privacy Act (or FERPA).12 Even universities like the University of California at Santa Barbara “have advised against the use of test proctoring services in remote learning” due to student protests with proctoring systems and the “aggressive tactics employed by proctoring companies in response to those [protests].”13
There has been concerns that the AI algorithm behind proctoring software are biased and discriminatory to some students. Simon Coghlan, Tim Miller, and Heannie Paterson from the Melbourne Law School, in their paper state that “the field of AI Ethics (and, more broadly, digital ethics) is young and still under development” and “[has] occasionally been criticized for their lack of practical specificity and theoretical philosophical rigor.”14 Despite these issues, proctoring companies advertise AI as a feature that improves academic integrity to universities on their sites and advertising campaigns. Their AI algorithms however, has raised the issue that the technology is not ready for today’s needs and are biased against minority groups in the same way as other AI attempts in the past, such as Amazon’s AI recruiting tool, which excluded applicants who used the words “women” and the names of all-women’s colleges in their resumes to work for the company.15
Senators Richard Blumenthal, Ron Wyden, and Chris Van Hollen in their letter to ExamSoft tell the company that “students of color, and students wearing religious dress, like headscarves, have reported issues with the software’s inability to recognize their facial features, temporarily barring them from accessing the software.”16 While it may seem like a technical limitation, the proctoring companies advertise this as “elegant, functional, [and] powerful” and “eliminates human error [and] bias” with no evidence, despite past incidents.17 Ian Linketter, a learning technology specialist, posted on Twitter his analysis of Proctorio’s system. In the post, he explains that Proctorio’s system was invasive of students’ privacy and claims that “it discriminates against students who are marginalized, neurodiverse, or who otherwise do not fit the developers’ definition of normal.”18 Nir Kshetri, a management professor at the University of North Carolina, Greensboro explains to The Conversation that “simpler algorithms such as [AI] applications have been mainly trained to identify white males and, consequently, misidentify ethnic minorities,” and believes that AI isn’t appropirate for remote proctoring today, which has caused many students and advocates to wonder whether what the proctoring software are claiming with their AI system really is fact.19
Proctoring companies will state that each AI algorithm they have is trained differently from others, and are each adapting to new behaviors and different races and ethnicities; however, Simon Coghlan, Tim Miller, and Jeannie Paterson’s paper shows that the bias and discrimination exists in their AI algorithm. In their paper, they state that AI “has been criticized as inaccurate, and has [resulted] in legal action, despite the fact that [it] may have been trained on thousands or millions of images.”20 Even then, the proctoring companies themselves don’t disclose the AI assessments they make to backup their claims of bias, which raises a concern as to how the proctoring companies advertise this feature without actual data and past incidents that shown how it treats students of different religions, races, or ethnicities. False flagging by the AI can “harm [student]s’ opportunities often in arbitrary and discriminatory ways,” and students have no access to see any flags made during the exam or why they were made.21 Students would be unable to figure out what they are being flagged for and why beforehand, which would make it difficult to prove their case to their university or to their professor if the professor thinks that they cheated on an exam in class.
Proctoring software also raises concerns over a student’s well-being during the exam. While many of us may be able to live life as everyone else, others unfortunately have conditions that are not always ideal for some students. No one knows about a student’s condition before the exam starts, whether it is the professor, a random proctor, or AI algorithm. Due to this, many students with conditions like ADHD, trichotillomania, and chronic tic disorder for instance, have a hard time trying to take exams with these systems, compared to taking it in person. Ketelyn Vercher, a sophomore at Georgia State University in Atlanta who has ADHD, states to The Signal that “using LockDown Browser while having unmedicated ADHD has made it hard for [her] to take tests.”22 Proctoring software can also harm students psychologically. Nir Kshetri explains to The Conversation that “one student vomited [during her exam] due to the stress” and had to do it “at her desk because no bathroom breaks were permitted” during her exam.23 Proctoring software not only harms students mentally, but also harms those that have health conditions from taking their exams. An AI algorithm or proctor won’t be able to recognize all the effects someone with mental stress or with ADHD may have and can determine the behavior as inappropriate, hurting the student even further for cheating accusations. Mental health and health conditions are real, and the way that proctoring software treats each student can causes more harm than it does good for a letter grade on an exam.
Universities and proctoring companies will state that universities can accommodate students with health concerns; however, the solution they make shows that proctoring software is not necessary to be used in class. At St. Mary’s University in San Antonio, Texas, students with health conditions are given a few choices depending on where they are located. In-town students can be proctored in a testing center while out-of-town students can be proctored by another university in-person, and out-of-country students can be proctored via Zoom. The accommodation, however, gets rid of proctoring software as a whole, which raises the point of using proctoring software in the first place, if these options get rid of the proctoring system as a whole for these students, but can’t be offered to other students. The accessibility services know these systems can treat students with health conditions poorly and the course of actions that they take to make sure they are taking the exam fairly should be considered by professors and universities to handle exams for others students that doesn’t harm the student’s well-being throughout the testing period.
Proctoring software raises concerns over how it stores students’ information. As mentioned previously, these proctoring companies like Respondus can store student data to be used in AI algorithm testing, but they also store other kinds of personal information, like student IDs or addresses. Nicholas Nguyen, a senior at San Jose State University used ProctorU to take his exam and told his university wire that “he was asked to answer four personal questions about himself,” one being the name of a family member to prove he was actually himself. It’s unusual that a ProctorU proctor would ask that kind of question of a student, especially for Nguyen, who had never told anyone about this particular information. To clarify the confusion, Franklin Hayes explains that ProctorU got this information from “public records and databases by Acxiom, a data-brokering company,” and that this method is used as a security measure by some credit card companies.24 While the information is public, it raises questions as to what these proctoring companies are gathering on students and why they need brokering companies to prove that the student is actually that student, compared to a student image and ID.
The student information held by online proctors, however, has to be stored somewhere. Normally, this data is stored on servers hosted by the company, and is secured to prevent unauthorized access to the system; however, these system can fail, which has caused personal information to be publicly revealed. Students who have used ProctorU have gotten their data leaked recently on a hacker forum. BleepingComputer’s creator Lawrence Abrams states that the data leaked contained “email addresses, full names, addresses, phone numbers, hashed passwords, the affiliated organization, and [more]” from the University of Texas, Harvard, Yale, and even universities in other countries like Australia.25 ProctorU remained silent about this breach until users on Twitter and journalists like Abrams started to report it, with the company confirming the breach on Twitter in early August 2020, after someone tweeted an article by the University of Sydney, Australia talking about the breach.
ProctorU stated in a update that “data collected by ProctorU prior to March 2015 appeared to have been acquired and posted by an unauthorized user.”26 However, Abrams’s in his article analyzed the leaked database and found that it contained student account information up to 2017.27 Not only has ProctorU remained silent about this up till August 2020, but they didn’t even check that the data had student information up to 2017. Even if the data is three to eight years old, a data breach of this kind is still something serious to take a look at, especially a breach that leaks students personal information. Data breaches do happen all the time, from Yahoo in 2014 to Equifax in 2017; but the existence of this kind of information shows how much data these companies retain from students, years after they stopped using their services. And those data breaches may expose many of these students to identity fraud due to the lack of security on the system’s servers.
Other proctoring companies will state that they don’t gather and store the same type of information like ProctorU does. This is untrue, however, since Jason Kelley and Lindsay Oliver from the EFF tells us that “[proctoring companies] retain much of what they gather[,] whether that’s documentation or video of bedroom scans,” and that they have “no time limits on [data] retention” and may share it to third-parties.28 Respondus, for example, claims that “the recordings are controlled by [the student’s] institution,” but admits that “random samples of video and/or audio recordings…may be shared with researchers (research institutions and/or biometric experts) under contract with Respondus,” which the EPIC calls “a transfer of data which would necessarily include identifiable images of students’ faces,” which misleads students and institutions as to what it shares to third-parties.29 Students don’t have control over how their data is handled, despite the fact that the data shows the students’ face in it or asks the proctoring companies to delete it, because according to them they aren’t considered the data’s ‘owner.’ Rather their institution does.30
Proctoring companies have leaked student data before. A student on the social platform Reddit complained about their experience with Proctorio’s support team when the software kept crashing on their midterms. Proctorio’s CEO Mike Olsen, who read the student’s post, decided to post the student’s support logs in public, telling the student, “If you’re gonna lie bro . . . don’t do it when the company clearly has an entire transcript of your conversation.” It was “quickly panned as an inappropriate use of corporate data and an invasion of privacy” by other users, and the CEO later apologized and removed the post.31 Using student logs like Proctorio’s CEO did on Reddit shows not only how the company stores their information, but how it uses that information against them. A company is supposed to keep this data private, not post it publicly when it’s criticized and when they can use this information against the student like this, it’s unprofessional, and a violation of student’s data rights and laws like FERPA.
With proctoring software causing more issues and concerns to students, professors, and activists than what it intends to fix, some universities are already looking at alternatives to handle their exams in different ways, with students leading the way with possible ideas. Alyssa Origer, a student at Delta College, suggests to professors that “if [they] want to make sure we aren’t cheating, they could just hold a Zoom meeting and monitor us themselves.” What this would do is allow professors to watch students take their exams just as if they were in-class and eliminate the issue of a random proctor or “a faulty [AI] algorithm deciding whether we are cheating or not.”32 Another solution is to make the tests and quizzes be short answer responses rather than multiple-choice. Making these kinds of exams allows professors to easily tell whether a student plagiarized their response, and makes the students put in actual effort for the exam topics to pass the exam. It won’t stop the use of Google and re-wording other people’s responses, but it can be easily shown whether or not a response was made with the use of the internet, someone else, or actual readings. One last option to consider is to assign a project to the class rather than a exam. Similar to short-answer questions, this solution will allow professors to check the quality of the work for any plagiarism with tools like Turnitin and make students put effort into the topic and the requirements the project needs. The professor can then review their work and plagiarism score and email students questions about their project to see whether the student had a understanding of the topic or violated academic honesty by using someone else’s work or the internet.
While the use of proctoring software is valid in today’s society and academic need, it is simply too much of a risk to students’ data and well-being than it is to maintain academic honesty in the classroom. “No student should be forced to make the choice to either hand over their [data] and be surveilled continuously or to fail their class.”33 Proctoring companies advertise that they prevent cheating, but a report by Michael N. Karim, Samuel E. Kaminsky, and Tara S. Behrend, at George Washington University in Washington, DC concluded that while it prevents cheating, “neither of the main effects were significant, suggesting there was no overall mean difference between the proctored and unproctored [exams].”34 Universities and schools that use proctoring software need to reconsider the options available to them and switch away from using such systems for exams and quizzes with something that can benefit, and be accessible to students more openly than a low cost fix for all exams that causes students to give up their data and risk their mental and general health.35 Until then, our current classes will still have to run these invasive proctoring systems that bring more negatives than positives to students over a simple letter grade in class for their future degree.
Azariel Del Carmen is a student majoring in Computer Science with a minor in Mathematics and has an interest in Cybersecurity for the 2024 school year at St. Mary’s University of San Antonio. Born in Ventura, California but raised in San Antonio, Texas, he loves to learn and hear new things relating to technology and help others out as best he can with their issues to make them happier. He is inspired by what he does, others he looks up to, and the person he wants to become for the future in his life and goals.Author Portfolio Page