Case studies
Real outcomes from real creators — anonymized to archetype, platform, and content type. Each card shows the situation, the outcome stat, and the timeframe, so you can read what the agent actually files against and how fast the platforms act. The full leaderboard of platform + content-type aggregates is on /leaderboard.
Want the cross-platform detection write-up before reading the stories? See how the agent works → /how-it-works
Recovery stories
Each story carries the creator archetype (live streamer, fitness educator, podcast host, independent news creator, music producer), the platform where the theft occurred, the content type (deepfake, partial clip, VOD rip, audio rip), a 2-3 sentence situation, the outcome stat, and the timeframe. None of these stories contains a creator name, a handle, a business name, a specific URL, an exact date, or a per-creator financial number.
"Removed in 14 hours"
Q1 2026
Three weekly broadcasts were lifted in full within hours of airing and rehosted to a clip-channel mirror and a Telegram audio-drop channel. The agent caught two on the first hourly sweep and the third within six hours — naming the publishing channel on each filing, not just the URL.
"Removed in 36 hours"
Q4 2025
A face-replica overlay paired with a voice clone of the on-camera instruction was cycling across a face-filter chain and a vocal-stub voice-note bot. The agent filed NO FAKES-aligned takedowns naming the publishing handles; both surfaces voluntarily removed the rip within the 48-hour window, and the second refusal was escalated to a TAKE IT DOWN FTC referral.
"Median time-to-removal under 11 hours"
Q1 2026
Weekly long-form episodes were being rehosted as audio-only drops on a file-host mirror chain within an hour of RSS publication, with the lifted intro clip also cycling into Telegram audio-drop channels. Chromaprint on the audio track caught every lift; the agent filed audio-channel takedowns naming both the publishing handles and the file-host abuse channels.
"Removed in 22 hours"
Q2 2026
Investigative-report segment clips were being lifted, re-titled with a misleading summary line, and fed to a Telegram forward-chain. Three-signal matching (perceptual + chromaprint + face/voice similarity) caught the partial clips even after re-encode; the agent filed against the publishing channels and a whitelist exception prevented a single false positive.
"Removed in 3 days · FTC escalation honored"
Q1 2026
A voice clone of an unreleased vocal excerpt was being paired with a face-replica overlay on a ghost-account clone handle and a voice-clone audio-message bot. The fingerprint + face/voice similarity stack ran clean; the agent filed against both handles, the publishing account refused, and the agent re-filed within the TAKE IT DOWN Act 48-hour window as an FTC referral that was honored on day three.
Want the platform + content-type aggregate breakdown? That lives on /leaderboard. Want the per-archetype platform notes? That lives on /how-it-works.
Anonymization
The stories above intentionally carry no creator name, no channel handle, no business name, no email or phone, no specific publishing-channel URL, no specific infringing URL, no exact date, and no per-creator financial figure. The only signals that leave the row boundary are archetype labels, platform names, content-type labels, narrativized situation copy, outcome stats, and timeframe ranges like “Q1 2026”.
Revenue impact is surfaced only as platform + content-type aggregates on /leaderboard, and the recovery digest masks individual creators to first name + last initial before anything leaves the inbox.
Turn the agent on
$349 / month. Free until the first takedown lands. Hourly scans across YouTube / TikTok / Instagram / Telegram + file hosts. NO FAKES-aligned likeness + voice takedowns. TAKE IT DOWN Act 48-hour mandate with FTC referral on refusal.
Last updated 2026-08-01. Stories are anonymized at write — no creator name, handle, business name, email or phone, specific URL, exact date, or per-creator financial figure appears on this surface.