What Honorlock Records in a Hiring Assessment

Honorlock is an exam proctoring company that spent a decade watching university students, and on 22 September 2026 it announced products aimed at employers instead. If a hiring team sends you a Honorlock-proctored interview or assessment, the monitoring you get is the education stack repointed at candidates: a Chrome extension that records your webcam, screen, and audio; an optional desktop application installed with administrator privileges that blocks unauthorized desktop apps; an ID and workspace check; and a human proctor who can interrupt you mid-session.

This article sets out what Honorlock's own documentation says it captures and flags, where its corporate claims and its student-facing privacy language sit awkwardly together, and what a candidate can reasonably conclude. Facts are current as of 26 September 2026 and come from Honorlock's published pages and its announcement, not from firsthand testing of a hiring deployment.

The hiring products are new; the monitoring is not

Honorlock's 22 September 2026 announcement describes four functions: securing remote interviews by "verifying candidate identity, blocking unauthorized AI and remote access tools and detecting suspicious activity during interviews"; protecting assessments against proxy participation and unauthorized AI; evidence documentation, meaning "recording session activity and providing timestamped reports and flags so hiring teams can review potential misconduct"; and a candidate experience that runs "in the background after identity and workspace verification without unnecessary alerts or interruptions" (Honorlock announcement, 22 September 2026).

The announcement cites a 2026 Greenhouse figure that 91% of recruiters and hiring managers have encountered deceptive candidate behavior. Greenhouse's own writeup of that number puts it as "91% of recruiters have already run into some form of candidate fraud," attributed to its 2026 AI in Hiring Report, and does not publish the sample size or methodology alongside it (Greenhouse on AI candidate screening and hiring fraud). Treat it as a vendor-surveyed sentiment figure that explains why employers are buying, not as a measured fraud rate.

Honorlock's corporate page is more concrete about the use cases it is selling into. Under talent acquisition it offers to "secure remote interviews and technical assessments while preventing proxy candidates and fraud in real time," and names anti-proxy identity verification, AI blocking, cell phone detection, BrowserGuard, and Search & Destroy as the components (Honorlock corporate and certification page).

What is recorded

Webcam, screen, and audio in one stream

Honorlock's student and faculty FAQ states that "if requested by your organization, we will collect a webcam video recording that includes your desktop activity and audio recording," and that the system monitors "the exam and web pages visited during the examination" (Honorlock faculty and student FAQ). Three things are worth separating out of that sentence. The capture is conditional on the organization's settings, so a hiring team can run a lighter configuration than a licensure board. The screen recording is your desktop activity, not only the assessment tab. And the three signals arrive as one reviewable artifact, which is what makes a timestamped report possible later.

Audio is transcribed, and specific phrases summon a person

Honorlock "records and transcribes audio while the exam is in progress," and its voice detection listens for a keyword list rather than analyzing speech generally. The documented examples are voice assistant wake words such as "Hey Siri" or "OK Google," the list "can be added to and modified by the institution," and when the system "detects phrases to use an AI assistant or other misconduct, a live remote proctor is alerted and can intervene" (Honorlock voice detection).

The practical reading: a customizable keyword list means the trigger set in a hiring deployment is not knowable from Honorlock's public pages. Honorlock says its approach "allows test takers to read aloud without being flagged for misconduct," which is a design goal rather than a published false-positive rate. Reading a problem statement aloud is documented as tolerated. Saying a wake word is not.

An ID photo and a workspace scan, run from your phone

Verification is a roughly 60-second "scan-and-go" process through Honorlock's mobile application: it logs a photo of your ID to check the name against the expected test taker, then walks you through a scan of your workspace (Honorlock ID verification and authentication). The FAQ adds that the room scan may be required before the session begins, and that if "no face is detected, or if multiple faces are detected, AI will flag the incident and a live human proctor may intervene."

The scan records whatever is in the room at that moment, not only your desk, and it is captured through a device Honorlock's own FAQ says it cannot otherwise access or control.

Retention is a default, not a rule

"The standard data retention period is 365 days, but organizations can adjust it to be shorter or longer," and after a session completes "our human proctors review the sessions" and produce reports that highlight incidents with timestamps (Honorlock faculty and student FAQ). For a hiring assessment that means the recording of your desk, your face, and your screen may outlive the decision on your application by a year, held by a vendor you never contracted with. The employer sets that number, so it is a fair question to ask the recruiter.

What is blocked and detected

BrowserGuard restricts the session around your browser

Honorlock describes BrowserGuard as restricting "access to other websites, browsers, and applications," disabling "keyboard shortcuts and functions," removing toolbar and menu options, and blocking and flagging "the ability to minimize the browser display screen," while permitting whitelisted resources (Honorlock BrowserGuard). Honorlock's own framing is that "browser lockdown software isn't enough," which is the honest part of the pitch and explains the next component.

The desktop application is the part that changes the threat model

Honorlock's help center documents a separate desktop application that works "in tandem with the Honorlock Chrome Extension," relaunching Chrome "in a protected state." It runs on Windows and macOS, is Chrome-only and explicitly not compatible with Edge, cannot be installed on Chromebooks, and "requires administrator privileges to install." Its stated purpose includes blocking "AI & cheating tools" by preventing "unauthorized desktop apps such as AI coding assistants," with administrator whitelisting available (Honorlock Application help center article).

That is a categorical difference from a browser recorder, and candidates repeatedly get it wrong. A browser extension observes the page and the tab. A locally installed agent with administrator rights operates at the level where desktop software lives. Honorlock's public documentation does not describe the mechanism, whether it enumerates running processes, what its block list contains, or how it handles virtual machines, so the precise capability is unverified from the outside. What is documented is the intent to block unauthorized desktop applications, and that intent is enough to plan around. The same distinction is worked through in what changes when a coding assessment ships as an app rather than a website.

The announcement also lists "blocking unauthorized AI and remote access tools," which puts remote desktop and screen-sharing software in scope. If you were considering running the assessment from one machine and working on another, how interviews and assessments surface remote desktop sessions is the relevant read, and the short version is that this is a well-covered detection target rather than a blind spot.

Cell phone detection and leaked question sweeps

Cell phone detection is a named feature on Honorlock's site and corporate page, and the FAQ is candid about the boundary: Honorlock "cannot access or control secondary devices, such as mobile phones." The detection is therefore inference from what the webcam and microphone observe, not instrumentation of the phone itself.

Search & Destroy is the feature candidates rarely hear about and should. Honorlock says it "finds your exam questions online and gets them removed" on the client's behalf. If you are preparing from a question bank that circulates on a forum or a paid dump site, the assessment you sit may be a rotated version of it, and matching answers to withdrawn content is the kind of pattern an integrity review notices.

The identity claim worth reading twice

Honorlock's corporate page promises that "AI-powered authentication confirms the person you interviewed is the same person who shows up on day one." Its ID verification page, written for the education market, states plainly that "Honorlock does not use face recognition, fingerprints, voiceprints, or other biometric technologies to identify test takers," and that it does "NOT identify any individual test taker's face, store facial elements, or match the face to a database."

Both statements can be literally true — name matching against a logged ID photo plus human review can establish continuity without a biometric template — but the marketing sentence implies a cross-session identity guarantee that the disclaimed technology set does not obviously deliver. This is not evidence of a contradiction; the corporate product may be configured differently from the education one. It is a specific question worth putting to an employer in writing, particularly under state biometric privacy statutes: whether a facial template is created, and if so, under what notice and consent. A candidate's general leverage here is covered in the candidate's guide to interview identity verification.

What a flag actually does

Honorlock does not adjudicate. It produces timestamped incident reports that human proctors have reviewed, and the hiring team decides. That matters because the failure mode most candidates worry about — a single ambiguous flag ending a candidacy silently — depends entirely on employer process, not on the proctoring vendor's thresholds. What typically happens after an assessment flags you sets out the realistic range of outcomes and what you can ask for.

Two specific consequences follow from Honorlock's design. A live proctor can intervene during the session, so a flag is not necessarily a silent post-hoc finding; you may get a pop-in and a chance to respond in the moment. And because reports are timestamped against a recording, a disputed flag has a reviewable artifact behind it, which cuts in both directions.

What this means before you sit one

Read the invitation for the install requirement. A Chrome extension alone means browser-scoped monitoring plus webcam, screen, and audio recording. An additional desktop application requiring administrator rights means a locally installed agent whose stated purpose includes blocking unauthorized desktop software, and you are consenting to that at install time. Those are different situations and the invitation usually distinguishes them.

Ask about retention and access. The 365-day default is adjustable by the employer, and asking which value they set is a reasonable, low-friction question that also tells you how carefully they have configured the deployment.

Do not plan around defeating the desktop agent. Control is a desktop-native AI interview assistant for interviews, assessments, and screen-share workflows, and Control's own documentation is specific about what that means: a native desktop overlay rather than a browser extension, which it states does not appear in supported screen captures and does not take focus from the window being shared. That claim is about capture and focus behavior. It is not a claim about an administrator-installed process monitor whose documented purpose is to block unauthorized desktop applications, and Control makes no such claim. A Honorlock deployment that includes the desktop application is a lockdown environment, and the honest answer for that environment is that no overlay tool should be assumed to survive it.

Where Control's category does fit is the much larger set of interviews that are ordinary video calls and screen shares with no installed agent. The same vendor sells very different configurations to different buyers, so the relevant question is never "is this Honorlock" but "which components did this employer turn on."

The short version

A Honorlock-proctored hiring assessment records your webcam, your desktop activity, and transcribed audio, checks your ID photo and your room through your phone, listens for a customizable keyword list that can summon a live proctor, restricts your browser, and — if the employer enables the desktop application — installs an administrator-privileged agent whose stated job includes blocking unauthorized desktop AI tools. The recording may be kept for a year by default, and the flags it produces are handed to the employer rather than acted on by Honorlock.

Two things remain unverified from public documentation: how the desktop agent detects what it blocks, and whether the corporate anti-proxy identity product creates a biometric template that the education-facing page disclaims. Both are worth asking about directly.

If your upcoming interviews are ordinary screen shares rather than agent-locked assessments, Control's focus test lets you check capture and focus behavior on your own machine before you rely on any assumption about it.