Roadmap
What we're building next
Concrete milestones for DeadMark and DeadDetect, focused on reliability, survivability, and practical integrations.
Now
Core media + text/code expansion (started early)
Current
- Tune image and video watermark strength by region to keep it invisible and recoverable.
- Tune audio watermarking by time window and frequency band.
- Expand transform tests (re-encode, resize, crop, compression, clipping, noise) with recovery scoring.
- Improve evidence views for image, video, and audio review.
- Text/document/code watermark preview tracks are already active in product UI.
- Deep Scan no-watermark AI-origin fingerprint outputs are now in active development.
In progress
AI-origin model rollout (no watermark path)
Current development track
- Train and evaluate attribution models for AI-origin estimation when no watermark is present.
- Surface media fingerprint-style evidence outputs (for example: DALL-E/Kling/Seedance-style model-family signals).
- Combine AI-origin confidence with DeadDetect evidence package and policy-ready output.
- Roll out in stages based on reliability thresholds and partner validation.
Next
DeadMark + DeadDetect pilot hardening
Q2 2026
- Stabilize encode/decode logs, UID trace outputs, and media-library audit flows.
- Ship cleaner integration examples for image, video, and audio pipelines.
- Improve survivability reporting for pilot evaluations.
- Expand backend integration docs for moderation and provenance workflows.
- Harden text/document/code verification and messaging-surface survivability benchmarks.
Later
DeadDetect API scale-up and enterprise integrations
Post pilot validation
- Production API workflows for verification, evidence retrieval, and audit logging.
- Batch and async scanning for moderation/compliance pipelines.
- C2PA + internal UID/log reconciliation for mixed provenance environments.
- Enterprise deployment options with reliability monitoring.