What to Do If the Other Side Says Your Evidence Is a Deepfake

Saying “that’s a deepfake” is now an easy thing to say about a video, photo, voice note or screenshot that hurts someone’s case. Courts are still working out how to handle it, but the practical answer for the person relying on the file is the same: show where the file came from, show it still matches its fingerprint, and back it up with other evidence. Here’s how.

Disclosure: I’m Inna Hall, co-founder of TimeBinder, which is mentioned in step 4. This page is general information, not legal advice. If you’re in a court case, follow your lawyer’s instructions on collecting and handling evidence.

Where courts are right now

In the US, the federal Advisory Committee on Evidence Rules met in Boston on 15 October 2026 with a draft deepfake rule (a proposed Rule 901(c)) and a revised Rule 707 on machine-generated evidence on the agenda. It adopted neither and set up a subcommittee, so any new federal rule is still years away. A Federal Judicial Center survey found that only 15 of 931 judges who responded had seen a deepfake challenge.

Until then, the existing rules apply. The person offering a file has to show it is what they say it is, and that bar is fairly low. Some courts have treated a bare, speculative “it might be AI” objection as something that goes to how much weight the evidence gets, not whether it’s let in. A specific challenge with real reasons is different and can need a real answer. Our overview of deepfake evidence rules covers the proposals in more detail.

Why “run it through a detector” isn’t the answer

Detection tools give scores, not answers, and they disagree with each other. At the October meeting, OpenAI’s policy counsel said she didn’t think there would be a scientific way to detect AI media going forward, and Meta’s representative said recent AI media can’t be told apart by people. In one study of 1,783 participants, people judged real and fake media correctly 47.4% of the time.

So the stronger answer is about the file’s history: where it came from, who handled it, and whether it has changed.

Step by step: making your files easy to trust

1. Keep the original, not a copy of a copy. The file from the camera, phone or recorder carries the most information. Screenshots, screen recordings and files saved from a messaging app have been re-encoded and lost metadata. Don’t edit, crop or “enhance” the original. Work on duplicates.

2. Use the platform’s own export where you can. For messages, an official chat export or account data download (such as WhatsApp’s export or Google Takeout) is better than screenshots. Save the email or download page that delivered it.

3. Write down how you got it. A short note: which device, who recorded it, when and where, who has had the file since, and where it’s stored. This is the start of a chain of custody, and it’s much easier to write on day one than to rebuild a year later.

4. Fix the file’s fingerprint at collection. A SHA-256 hash identifies the exact bytes. Timestamping it gives you verifiable evidence of when the file existed, so you can show it existed no later than a date before the dispute arose and still matches its fingerprint. With TimeBinder, files are hashed in your browser at app.timebinder.io and never uploaded, which matters for sensitive material. Up to 10 files fit in one TimeBind, and you can zip a larger set and keep that exact zip.

5. Keep what surrounds it. Other people’s copies, the original message thread, location history, call logs, receipts, and witnesses who were there. Evidence that fits with other evidence is much harder to call fake.

6. Get an expert when it matters. If the challenge is specific and the file is central, a forensic examiner can look at the device, the file structure and the metadata. They’ll want the original and your notes from step 3.

If you’re the one questioning a file, the same list works in reverse: ask for the original, its source device, how it was collected, and any hash or timestamp made at the time.

Which plan fits

The free plan gives you 1 Token a month, which is one Pooled TimeBind of up to 10 files. That’s enough for a small set of evidence.

If you collect files often, Essential is US$20 a month for 5 Tokens (5 Pooled, or 1 ASAP plus 2 Pooled). Expert is US$50 a month for 15 Tokens (15 Pooled or 5 ASAP), and unused Tokens roll over. Pooled is usually recorded within 24 to 48 hours, and often sooner, depending on Bitcoin network activity. ASAP costs 3 Tokens and gets its own transaction (ASAP vs Pooled). Verification is free. See pricing.

What a timestamp does and doesn’t do

TimeBinder anchors each file’s SHA-256 fingerprint to Bitcoin and gives you a Proof of Time Certificate. Anyone can click Verify File and check, in their own browser, that the file existed no later than the anchor date and still matches its fingerprint (how verification works). Our evidence authentication explainer covers how this fits with other records.

It doesn’t show that the content is real, who recorded it, or what happened before it was timestamped. A fake made last year and timestamped last year would still carry last year’s date. Whether a court accepts any piece of evidence is up to the court. What a timestamp does is rule out “it was made after the dispute started” and “it’s been altered since”, and it’s a small, cheap step to take when you first collect a file.

Plan details were checked against timebinder.io/pricing on 28 October 2026.