Frequently asked questions
Frequently asked questions
Direct answers regarding forensic accuracy, multi-frame video decomposition, data privacy, and verification credit packs.
Forensic capabilities & accuracy
What technologies power Sealed Rose’s AI detection?
Sealed Rose combines foundation classification by our detection partners with our own multi-frame video decomposition across 10–45 sampled frames, Short-Time Fourier Transform (STFT) acoustic spectrogram analysis, and spatial diffusion fingerprinting to detect generative artifacts.
Which video generation models can you detect?
Our models detect leading generative video architectures including ByteDance Seedance 2, Wan 2.1, Kling, OpenAI Sora, Luma Dream Machine, and Runway Gen-3, as well as real-time face-swap blends.
Can you detect cloned voices and audio deepfakes?
Our STFT acoustic engine isolates robotic phase continuity, harmonic pitch flattening, and neural vocoder anomalies characteristic of voice cloning models such as ElevenLabs.
Can I scan videos directly from social media links?
You can paste public URLs directly from YouTube, TikTok, Instagram, X (Twitter), Facebook, or Reddit without downloading the video first. Our forensic resolver extracts the direct stream and decomposes frames for temporal neural scoring.
What video and image formats and file sizes are supported?
For direct file uploads, we support MP4, MOV, WebM, and AVI video containers up to 50 MB (up to 5 minutes), as well as JPG, PNG, WebP, AVIF, HEIC, and GIF images up to 10 MB. All analysis uses hardware-accelerated decoding.
How does Reverse Face Search and Likeness Defense work?
We extract 512-dimensional facial landmark vector embeddings to search public web images via reverse image search for unauthorized deepfakes and catfish impersonation without storing raw biometric imagery.
Privacy & ephemeral processing
Is my uploaded media stored or used to train AI models?
Raw media is held only in RAM during the scan and deleted right after. All in-memory stream chunks and processing buffers are freed immediately. We never train public or private foundation models on customer files.
How do you prevent Child Sexual Abuse Material (CSAM)?
Every uploaded image and video frame is hashed against Microsoft PhotoDNA before forensic scoring. Any match triggers immediate blocking, evidentiary quarantine under 18 U.S.C. § 2258A (REPORT Act), and an automatic report to NCMEC.
How do I remove an unauthorized deepfake or intimate image of myself?
Submit a request through our no-login removal portal at /report/ncii under the TAKE IT DOWN Act. Verified depictions are removed within 48 hours and perceptual hashes are blocklisted to prevent re-indexing.
Pricing & credit packs
Do I need a monthly subscription?
Sealed Rose provides free daily allowances across all verification tools: 1 free check a day without registration (1 bonus check with a free account), with zero recurring subscriptions. High-volume analysis is powered by one-time, non-subscription Verification Credit Packs that never expire.
How are credits consumed?
1 credit per image, audio clip, text check, or phone lookup; 12 credits per deepfake video scan (accounting for multi-frame neural decoding and temporal model scoring).
What happens if a scan is inconclusive?
We operate an automatic Grace Retry Policy: if an upload or stream yields an inconclusive verdict due to severe compression or resolution limitations, that check is forgiven and does not consume your daily allowance.
Do credits expire?
Verification credits remain in your balance indefinitely with zero maintenance fees.
What is your refund policy?
Unconsumed Verification Credit Packs are 100% refundable within fourteen (14) calendar days of purchase directly through Stripe.
Still have questions?
Our forensics and security engineering team responds within ≤24 business hours.