How to Check a Suspected Deepfake Video: Five Practical Steps

A suspicious video should be checked against its original source and the event it claims to show, not judged only by a face or an automated detector score. The five-step workflow below combines source tracing, context checks, frame searches and audio review. Until the claim can be verified, avoid forwarding the clip as fact.

What a deepfake check can and cannot establish

A misleading clip may be AI-generated, edited, dubbed, or genuine footage attached to a false date or claim. Those are different problems. A reverse-image match can help trace earlier footage, but a missing match does not prove that a video is new or authentic. Equally, a strange blink is a reason to investigate, not a verdict.

Australia's eSafety Commissioner guidance recommends checking context, changes in lighting or backgrounds, unnatural movement and mismatched sound, alongside reverse-image searching. Use these clues together rather than treating one imperfection as a reliable test.

Five steps before you believe or share the clip

1. Find the original upload. Open the post rather than relying on a screenshot or forwarded copy. Record the account, caption, upload time and link. If it claims to show an official announcement, look for the full statement on the institution's own website or established account. Do not assume a familiar profile image proves identity.
2. Check the event, place and date. Write down what the caption actually claims. Search for the named event and compare independent coverage. A genuine video from a different year can mislead without any synthetic face or voice.
3. Search a clear frame. Pause on a distinctive scene, save a screenshot and try a reverse-image search. Compare earlier uses and their captions. Look for a longer recording that preserves what happened before and after the edited segment.
4. Review picture and sound together. Watch the mouth, face edges, background and lighting, then listen for abrupt changes or sound that does not fit the visible speech. Compression, poor connections and ordinary editing can also introduce glitches. Do not declare a clip fake solely because it looks untidy.
5. Verify the request through a separate channel. If the clip or voice asks for money, account access or an urgent action, stop. Contact the person or organisation using a number you already trust, not one supplied in the suspicious message. An apparent voice or face is not sufficient authorisation for a payment.

If verification remains inconclusive

Describe the clip as unverified, save the evidence and avoid repeating the allegation as established fact. Do not upload private recordings or identity documents to an unknown detection website. A detector result can be an additional lead, but should not replace checking the original recording and the underlying claim.

Reporting and evidence in India

Keep the original URL, account handle, screenshots and relevant transaction details. Use the platform's reporting route for impersonation or manipulated media. For cybercrime complaints in India, consult the National Cyber Crime Reporting Portal. If money has been lost, contact the bank or payment provider promptly and seek assistance on 1930. A Home Ministry reply identifies the portal and helpline; reporting does not guarantee recovery or immediate removal.

Sources and scope

This is a practical verification guide, not a forensic examination of a particular video. The source guidance linked above supports the checking and reporting steps. The featured picture is an illustration, not evidence that any depicted face or person participated in a real deepfake incident.