Trust & Transparency

Responsible Use Policy

AI Video Detector is a review tool — not a judge. Use detection results as one input among many when evaluating video authenticity.

Core principles

Results are probabilistic, not proof

A high AI-likelihood score means the scanner found suspicious signals — it does not prove the video is fake. A low score means few signals were found — it does not prove the video is real. Never use a scan result as the sole basis for accusing someone of deception.

Combine with source verification

Always check the source of the video: who posted it, when, where, and whether they have a track record of authentic content. A scan result without source context is incomplete.

Respect privacy and consent

Only upload videos you have the right to analyze. Do not upload videos of people without considering their privacy. Do not use the tool to harass, stalk, or target individuals.

Consider the consequences of false accusations

Labeling a real video as AI-generated (a false positive) can harm reputations, spread misinformation, and undermine trust. Consider the impact before sharing scan results publicly.

Do

Use scan results as a first-pass screening signal to decide whether deeper verification is needed.
Review evidence frames and reason codes before drawing conclusions.
Cross-reference with the video source, creator reputation, and other verification methods.
Report suspected deepfakes to the relevant platform using their official reporting channels.
Cite the scan as one piece of evidence, not as a definitive verdict, when communicating findings.

Do not

Treat a high score as forensic or legal proof that a video is fake.
Use scan results to publicly accuse someone of creating or distributing deepfakes without additional evidence.
Upload content you do not have the right to submit or analyze.
Use the tool to target, harass, or intimidate individuals.
Assume a low score means a video is safe or authentic — sophisticated AI output can produce low scores.

For specific use cases

Journalists and newsrooms

Use the scan as triage — not as a replacement for source verification, fact-checking, or expert review. A high score justifies deeper investigation, not publication of an accusation.

Trust and safety teams

Use scan results as one signal in your moderation workflow. Combine with user reports, account history, metadata analysis, and platform-specific context before taking action.

Educators and researchers

The tool can help illustrate AI detection concepts, but results should be presented with full context about limitations. Do not cite scan results as research evidence without independent verification.

Individual users

Before sharing a suspicious video, check the source. If the video comes from an anonymous account, has no clear origin, or makes extraordinary claims, treat it with skepticism regardless of the scan result.