AI Overlay for Respondus Monitor (vs Plain LDB)

Respondus Monitor adds webcam and screen recording on top of LockDown Browser. The recording is uploaded to Respondus' cloud and reviewed asynchronously by ML plus an optional human reviewer. Plain LockDown Browser without Monitor only locks the system; Monitor is the layer that records.

Respondus Monitor is the recording layer; LockDown Browser is the lockdown layer. They run together but are technically separate. Monitor uses standard macOS capture APIs because that's what Apple ships - there is no kernel extension or system extension required. ScreenCaptureKit on Mac honors window privacy at the OS level, before the captured frames reach Monitor. LDBypass sets the flag, so the frames Monitor uploads are identical to what they would be without the overlay running. Monitor's ML reviewers and the human spot-checks see the unfiltered exam screen but not the overlay region.

┌── macOS host ─────────────────────────────────┐
│  ┌── proctor desktop app ───────┐             │
│  │  Calls ScreenCaptureKit /    │  ← misses   │
│  │  CGWindowListCreateImage     │    flagged  │
│  └──────────────────────────────┘    windows  │
│                                               │
│  ┌── LDBypass overlay window ───┐             │
│  │  sharingType = .none         │ ← invisible │
│  └──────────────────────────────┘   to capture│
└───────────────────────────────────────────────┘