Proctoring companies like Honorlock that track students' keystrokes and cursor movements during exams to prevent cheating allegedly punish normal behavior
An unsettling glimpse at the digitization of education. — A Florida teenager taking a biology class at a community college got an upsetting note this year. Tweets: @zeynep , @random_walker , @pt , @_aloisi , @hypervisible , @autumm , @dealbook , @nytimesbusiness , @nytimestech , @nytimestech , and @evanselinger Tweets: @zeynep : Always read @kashhill. Many institutions use automated surveillance to decide if a student is cheating on tests as part of remote learning. Often, this means a clunky, unaccountable algorithm barely overseen by low-paid workers abroad has great power. https://www.nytimes.com/... Arvind Narayanan / @random_walker : A recurring pattern in AI that judges people: -The vendor markets it as a one-stop solution to win clients. -But the fine print puts responsibility for oversight on the client (company/school/gov't). -When the tool inevitably makes mistakes, both parties say it's not their fault. https://twitter.com/... https://twitter.com/... Parker / @pt : I, for one, an outraged that software to detect cheating would be improved by using example data in which humans determine students are cheating. https://twitter.com/... Antonio Aloisi / @_aloisi : The pandemic was a boom time for companies that remotely monitor test takers bringing #surveillance into academic settings. Some have been frustrated by the invasiveness, glitches, false allegations of cheating & failure to work equally well for all people https://www.nytimes.com/... @hypervisible : So many important elements in this NYT story, but two I want to highlight: Students not only have this system used against them, they are also used as data to “improve” the tech. Not a new thing, but it's always invasive and gross. https://www.nytimes.com/... https://twitter.com/... @autumm : “I try to become like a mannequin during tests now,” Hey #edtech, what exactly are you teaching? https://www.nytimes.com/... @dealbook : “Schools seem to be treating it as the word of God,” Cooper Quintin of the @EFF said about anti-cheating software. “If the computer says you're cheating, you must be cheating.” https://www.nytimes.com/... @nytimesbusiness : The software said she might have cheated during her biology exam. The professor she'd never met agreed. The video showed her looking down. She said she was just thinking. How does someone prove they are innocent when the video is ambiguous? https://www.nytimes.com/... @nytimestech : Despite widespread criticism, online proctoring software is still used at many schools around the country. When this student in Florida was accused of cheating by an A.I. monitoring tool and a professor she had never met, her experience was Kafkaesque. https://www.nytimes.com/... @nytimestech : The software said she might have cheated during her biology exam. The professor she'd never met agreed. The video showed her looking down. She said she was just thinking. @kashhill reports on how sideways anti-cheating software can go. https://www.nytimes.com/... Evan Selinger / @evanselinger : “Normal behaviors are punished by this software.” Except that what counts as abnormal appears to be basic human stuff, the embodied cognition of moving your head + hands to think & remember. Great reporting from @kashhill. https://www.nytimes.com/...
Context & Ripple Effects
This closes a two-year arc that began when pandemic lockdowns made services like Examity standard exam infrastructure in 2020 (widespread pandemic adoption) and continued as Proctorio grew its client base more than 500% from 2019 to 2021 even while facing lawsuits and university pushback (complaints, lawsuits, and accusations of bias). The New York Times' Honorlock reporting adds a new layer: keystroke and cursor tracking that allegedly flags ordinary behavior as cheating.
The deeper significance is who holds power inside these systems — per the coverage, an 'unaccountable algorithm barely overseen by low-paid workers abroad' can decide a Florida community-college student's academic fate. The same institutions are also extending data collection past exams entirely, as one California college's use of homework trackers and license plate readers shows (campus-wide movement tracking).
First-order effects
- Students at schools using Honorlock face false cheating flags for normal typing and cursor behavior, with appeals routed through human reviewers who see only what the software surfaces.
- Institutions that adopted proctoring during the pandemic now own the reputational and legal exposure of vendor algorithms they do not control.
Second-order effects
- Universities weighing renewals must price in litigation and bias claims — yet Proctorio's post-controversy growth shows demand has so far punished critics more than vendors, weakening institutional bargaining power.
- Vendors compete on detection breadth rather than accuracy: facial detection, eye tracking, room scans, and now keystrokes become selling points even as each new signal multiplies false positives.
Third-order effects
- If the pattern holds, exam proctoring becomes the beachhead for continuous campus surveillance — the California college case suggests the same toolset migrating from test-day monitoring to everyday student analytics.
- Accountability structures lag deployment: automated judgments executed by outsourced, low-wage reviewers point toward a durable model of algorithmic enforcement without meaningful appeal, likely drawing eventual regulatory attention.
The trend: Remote proctoring is evolving from pandemic-era exam monitoring into always-on student surveillance, with algorithmic flagging outpacing any mechanism for students to contest it.