Does Detecto handle long-form video?

Yes. Video is analyzed file-based with a per-scene timeline, so you see where synthesis indicators concentrate instead of a single blended number. Larger files take longer to process.

Can I run a batch of incoming media through the API?

Batch and programmatic workflows are something we’re expanding. Tell us your volume and setup through contact and we’ll walk through what’s possible today.

What’s retained after a scan?

The report and the media are saved privately to your account so you can reopen them, never public unless you share. Run in ephemeral mode to store nothing. Account deletion permanently removes account-scoped data where possible once it completes.

For publishers

For publishers verifying citizen-contributed media

Per-scene timeline and metadata-integrity checks on the file before it lands in the article.

Citizen photos, field audio, and short video arrive faster than they can be vetted. Detecto checks the file itself, a per-scene timeline, synthesis indicators, and metadata integrity, and returns an AI-likelihood before the media lands in the article.

See sample report How calibration works

Built for the people who review other people’s content

Editors at independent magazines
University writing centers
Newsroom verification desks
PR review teams
Content QA leads at agencies

What you review

What publishers typically scan with Detecto.

Citizen-submitted photos and short videos
Audio interviews submitted from the field
Press-release supplied imagery
Stock-like media used in editorial composites

What’s at risk

The failure modes Detecto is built to surface.

A synthetic image or clip is submitted as on-the-ground footage
A real photo is altered or re-encoded to hide manipulation
Press-supplied imagery is partly AI-composited
A deepfaked voice passes as a recorded interview

Where Detecto fits

The workflow, end to end.

Five steps. None of them is automated.

01
Upload the file, image, audio, or video.
02
Read the overall likelihood and confidence band.
03
For video, scan the per-scene timeline to see where signals spike.
04
Check metadata integrity for signs of re-encoding or stripping.
05
Keep the report on file as the basis for running or holding the media.

What the report gives you

Evidence that goes in the file.

A 0–100 AI-likelihood with a High / Medium / Low confidence band, per modality.

A per-scene timeline for video, plus surface-artifact and metadata checks for images.

A private, timestamped record tied to the exact file you reviewed.

What not to do with AI detection

The same posture every Detecto reviewer reads. The score is evidence, not a verdict.

Don’t publish or spike a file on the score alone, corroborate the source.
Don’t read “Likely human” as “verified authentic”; it means no strong AI signal, not provenance.
Don’t treat metadata as conclusive, it can be stripped or forged.
Don’t drop the desk’s normal verification because a number looks clean.

All four modalities