New format option flags AI-generated uploads the moment encoding completes, before viewers see them.

LOS ANGELES — May 26, 2026 — Qencode today announced the general availability of AI Detection, a new capability that returns a synthetic-vs-real verdict on every video upload the moment encoding completes. The feature is available now to all Qencode customers and integrates into existing pipelines by adding a single output to the job’s output format array.

We built this feature in response to user feedback that manual review cannot keep pace with AI-generated UGC volume. Lower-traffic synthetic uploads frequently accumulate views without ever crossing the complaint threshold that triggers human review. AI-Detection closes that gap by producing a verdict for every encode, regardless of audience size or downstream traffic.

How it works

When users choose the AI Detection output, the video is analyzed and Qencode produces a JSON artifact with all the details needed to make a decision. The artifact carries a top-level verdict of AI-GENERATED or REAL, a video_ai_ratio confidence score, and a per-frame array containing an is_ai flag and reasoning text for each sampled frame.

Users can route detection results by confidence rating: high-confidence verdict to fast-track removal, mid-confidence calls to a human review lane with the model’s reasoning attached, and low-confidence calls through to catalog. Moderation policy thresholds remain under each customer’s control.

AI Detection works as a moderation triage signal, so that high-confidence calls clear at upload, while ambiguous cases can route to human reviewers who remain the quality gatekeepers. Override paths and audit trails stay with the customer’s moderation team. To get started, see the Detecting AI-Generated Content tutorial.