MediaLive adds A/B watermarking for live-video leak tracing

AWS announced A/B forensic watermarking support for AWS Elemental MediaLive on September 10, 2026. The feature produces paired live-video outputs that downstream packaging and CDN infrastructure can assemble into session-specific sequences to help trace unauthorized redistribution, according to the AWS announcement.

What MediaLive produces

AWS says a single MediaLive channel produces two synchronized output variants, A and B, each carrying a distinct, visually transparent watermark. The company says those watermarks persist through re-encoding and screen capture.

The session-specific sequence is assembled downstream, not by the encoder alone. Packaging and CDN infrastructure combines the variants into unique per-session sequences that AWS says can identify the origin of leaked content. Both the workflow and the watermark-durability claims come from the release notice.

Configuration and availability

Customers can configure watermarking on Common Media Application Format (CMAF) Ingest output groups through the MediaLive API or console. MediaLive delivers the A and B variants via CMAF ingest to AWS Elemental MediaPackage or third-party packagers. AWS names Amazon CloudFront as compatible CDN infrastructure for downstream per-session assembly.

AWS says the implementation follows the DASH Industry Forum A/B watermarking specification, ETSI TS 104 002, to support interoperability with standards-compliant packagers and CDN infrastructure. The company lists the feature as available in all AWS Regions where MediaLive is available. These configuration, compatibility and availability details are stated in the announcement.

What engineers should validate

The key integration boundary is between producing watermark variants and assigning them to sessions. In the announced architecture, enabling watermarking in MediaLive is only one part of the attribution workflow: downstream systems must preserve the A/B distinction and assemble the session-specific sequence.

Teams should validate that behavior across their chosen packager and CDN before treating the encoder setting as an end-to-end leak-tracing deployment. The announcement describes a tracing capability, not a guarantee that unauthorized copying can be prevented.