Proctoring & Violation Detection Agent
Camera-based integrity for exams and interviews — flags, never final verdicts.
What it does.
The Proctoring & Violation Detection Agent watches an authorized exam or interview stream with computer vision, detecting phones, headphones, second screens, absent or multiple faces, and gaze anomalies, and raising time-stamped flags a human reviewer adjudicates.
It runs only on streams you authorize, logs every detection for review, and never issues a verdict on its own — the reviewer decides what a flag means.
Capabilities.
Object detection
Detects phones, headphones, and second screens in frame with a modern vision model.
Presence & gaze
Flags absent, multiple, or substituted faces and off-screen gaze patterns.
Time-stamped flags
Raises reviewable, time-stamped events — not an automated pass/fail.
Authorized & logged
Runs only on authorized streams and logs every detection for audit.
FAQ
Frequently asked questions.
- What does the Proctoring & Violation Detection Agent do?
- The Proctoring & Violation Detection Agent watches an authorized exam or interview stream with computer vision, detecting phones, headphones, second screens, absent or multiple faces, and gaze anomalies, and raising time-stamped flags a human reviewer adjudicates. It runs only on streams you authorize, logs every detection for review, and never issues a verdict on its own — the reviewer decides what a flag means.
- Is the Proctoring & Violation Detection Agent governed and sovereign-deployable?
- Yes. The Proctoring & Violation Detection Agent runs through the AgentAnywhere governed gateway — policy enforced at the point of every call, every action logged and attributable, and a signed Trust Receipt available for audit. For vision & detection, that means it runs only on the streams you authorize, raises reviewable flags rather than verdicts, and logs every detection for a human to adjudicate. It deploys inside your own infrastructure and jurisdiction, on your keys, and keeps a human in the loop — it removes the routine work and leaves the decision with your team.
- How do I get access to the Proctoring & Violation Detection Agent?
- The Proctoring & Violation Detection Agent runs as part of an AgentAnywhere pilot. Request access from this page and a solutions architect will scope a six-to-ten-week pilot against one real workflow — starting with object detection — in your sovereign environment. Once you're onboarded, the Sign in action opens your agent's own console via per-agent single sign-on.
Put the Proctoring Agent to work.
Pilot it against one workflow in your sovereign environment — your data, your keys, your governance. Already onboarded? Sign in to your console.
Sign in opens your agent's console · access requests answered within one business day