AI Overlay for ProcMonitor / AI-Proctor on Mac
ProcMonitor (marketed at ai-proctor.com as "AI Proctor") is a niche AI-based proctoring product used mostly by technical training providers and smaller certification programs. It runs as a browser-launched lockdown layer that forces incognito full-screen mode and notifies the instructor in real time of any tab, app, or website change.
ProcMonitor architecture is small and explicit: the exam opens in incognito and full-screen, facial verification runs first against a stored ID photo, then a live dashboard lets the instructor watch for events. If the candidate opens any other website (even in a new tab or new window) ProcMonitor screenshots the activity and notifies the instructor; if they switch to an already-open program (a PDF reader, a chat app), the same. The instructor can chat in real time or kick the candidate out. There is no kernel driver, no kernel extension, and no native helper that bypasses Apple screen-capture rules; everything runs in the browser plus a server-side dashboard.
Key points
- ProcMonitor / AI Proctor is a smaller niche player aimed at technical training providers and corporate certification programs.
- Runs entirely in the browser: opens the exam in incognito and full-screen, forces re-entry to full-screen if the user escapes it.
- Real-time instructor dashboard receives screenshots of tab opens, new windows, and app switches; chat and kick controls included.
- Identity verification with a single stored photo before the exam starts; webcam pipeline is independent of screen capture.
- Offline mode encrypts and caches detected events locally, then uploads when internet returns; same browser-capture API is used.
How it works
┌── macOS host ─────────────────────────────────┐ │ ┌── Safari/Chrome ─────────────┐ │ │ │ ┌── proctoring extension ─┐ │ │ │ │ │ Browser screen-capture │ │ ← reads │ │ │ │ → exam recording │ │ visible │ │ │ └─────────────────────────┘ │ pixels │ │ └──────────────────────────────┘ │ │ │ │ ┌── LDBypass overlay ──────────┐ ← rendered │ │ │ ChatGPT / Claude / Ollama │ excluded │ │ └──────────────────────────────┘ from cap │ └───────────────────────────────────────────────┘
Compatibility on Mac
| macOS 14 Sonoma+ | Yes (required for window-privacy API) | ✓ |
| Apple Silicon (M1/M2/M3/M4) | Native arm64 build | ✓ |
| Intel Macs (2019+) | Universal binary supported | ✓ |
| ProcMonitor / AI-Proctor browser layer | Browser screen-capture honors window privacy | ✓ |
| Real-time instructor screenshots | Source frames filtered before send | ✓ |
| Offline-mode event cache | Replays same filtered captures | ✓ |
Common questions
Will the instructor get a real-time screenshot showing the overlay?
No. The screenshot is taken from the browser screen-capture API, which on Mac obeys window privacy. The overlay window is filtered out before the screenshot is produced.
Does ProcMonitor detect that I escaped full-screen by toggling the overlay?
The overlay does not change browser full-screen state. ProcMonitor full-screen enforcement therefore does not fire.
Will identity verification fail because the overlay is open?
No. Identity verification compares the live webcam frame to your stored photo. The webcam pipeline is separate from screen capture.
What if internet drops mid-exam? Does offline cache see the overlay?
No. The offline cache stores the same screenshots the live mode would have sent, captured through the same filtered API.
Is the overlay invisible to the instructor real-time chat view?
Yes. The chat view is a side panel; the exam view it sits next to is built from the same filtered screen-capture frames.