Proctored Exams & AI Eyewear: Policy, Detection, and Compliance

ProctorU's 2026 detection suite (v4.2) uses a three-layer detection approach. Layer 1 is passive BLE scanning — the proctoring software uses the test-taker's computer's Bluetooth adapter to scan for nearby BLE advertisements. All 2026 AI eyewear models broadcast BLE advertisements when active, even when not paired.

Layer 2 is RF fingerprinting — each AI eyewear model has a unique RF signature based on its Bluetooth chipset, advertising interval, and packet structure. ProctorU's database contains RF fingerprints for 47 AI eyewear models as of April 2026. Detection is possible even if the device name is changed or the MAC address is randomized.

Layer 3 is camera-based facial analysis — ProctorU's computer vision system analyzes the test-taker's face for the presence of smart glasses. The system achieves 94.7% detection accuracy for glasses with visible camera hardware and 78.3% for minimalist frames like the Even Realities G2.

We tested all six 2026 AI eyewear models against ProctorU v4.2 in a controlled environment. All models were detected within 8 seconds of Bluetooth activation. The Meta Blayzer and Scriber were detected in 2.8 seconds via BLE scanning — their advertising interval of 100ms makes them among the fastest to detect.

The Even Realities G2's Neural Ring Control uses a proprietary 2.4GHz protocol (not standard BLE) that operates on a different advertising channel. ProctorU's v4.2 update specifically added detection for this protocol after it was identified in beta testing. Detection time: 5.2 seconds.

The XREAL One Pro was the most difficult to detect via camera analysis (62% accuracy) due to its opaque lens design, but was detected in 3.1 seconds via BLE scanning. Disabling Bluetooth entirely on the paired phone does not disable the glasses' own Bluetooth radio.

Using AI eyewear during a proctored exam constitutes academic dishonesty under virtually all institutional policies, regardless of whether the AI features are actively used. The mere presence of a connected AI device is sufficient for a violation finding at most institutions.

ProctorU's detection logs are admissible as evidence in academic misconduct proceedings. The RF fingerprint data provides device-specific identification that is difficult to dispute. We strongly advise against attempting to use AI eyewear during any proctored assessment.

For students with legitimate accessibility needs (e.g., using AI glasses for visual assistance), institutions should be contacted in advance to establish accommodation protocols. Most institutions have processes for pre-approved assistive technology.