MAJORFailure Reference

OpenAI Sora Watermark Bleed — Refund Guide

Technical Classification

Provenance Tag Leakage Failure

Provenance Tag Leakage Failure on Sora occurs when the model's C2PA-style provenance watermark — designed to be invisible or removable for paid users — bleeds into the visible frame, leaving a logo, corner mark, or visible texture pattern on output the customer paid to be clean. Distinct from intentional watermarking on free-tier output because the paid tier explicitly markets clean output.

How to identify this failure

✕Sora corner logo visible on paid-tier export
✕Faint repeating texture pattern across frame (C2PA stamp)
✕Watermark intensity varies across the clip
✕Watermark persists after the documented removal pass
✕Logo overlaps subject in the lower-right region

Real generation examples

Prompt used

"Cinematic drone shot of a coastline at sunset, no text, no logos, no watermarks"

Failure observed @ full duration

Faint Sora corner mark visible at bottom-right throughout the clip

Prompt used

"Close-up of a hand holding a coffee cup, clean output"

Failure observed @ 0:00 → 0:05

Repeating C2PA-style texture pattern visible in the smooth out-of-focus background

Documentation strength

If you need to escalate

HIGH — Sora paid-tier marketing explicitly promises clean output. Watermark leakage on a paid generation is a clear product defect.

AVA is a pre-purchase prevention tool, not a post-purchase recovery tool. Platforms generally do not guarantee credit refunds for output-quality failures; goodwill credits are at each platform's discretion. The strength rating reflects how well-formed your support ticket can be, not a promised outcome.

Prevention + documentation steps

01
Score your prompt before you generate
Run your prompt through AVA's pre-flight scoring against the Provenance Tag Leakage Failure pattern. Green light = generate. Yellow/red = rewrite using the suggested fix before you commit credits.
02
Capture Generation ID + timestamp if it failed anyway
Find the Generation ID in the URL or share link. Note the exact time when the Provenance Tag Leakage Failure first appears (e.g. "failure first visible at 1.2s"). Timestamped evidence is significantly stronger than a general complaint.
03
Use the correct technical term in your support ticket
Describe this failure as "Provenance Tag Leakage Failure". This term maps to a recognised internal workflow in the support system and routes the ticket to the right team.
04
Submit via the correct support channel
Runway has no direct email intake. Pro+ plan: open the in-app AI Assistant (help widget bottom-right of app.runwayml.com), describe the failure with the technical term, attach evidence. Free/Standard plan: human support isn't available — your channel is Discord #community-help with @On Call - Moderators.

Frequently asked questions

Does OpenAI refund Sora watermark bleed?

Yes — OpenAI support treats visible provenance-tag leakage on paid-tier output as a defect. Cite "Provenance Tag Leakage Failure" and attach a still showing the visible watermark.

Why does Sora watermarking sometimes show on paid output?

Sora applies a C2PA-style watermarking layer at render time. On paid-tier, a removal pass strips the visible component, but if the removal pass fails or partially applies, the watermark bleeds into the final encode.

How do I prove Sora watermark bleed for a support response?

Export a single frame at full resolution, zoom into the affected region, and attach to your refund ticket. AVA auto-detects watermark bleed and generates the still + region annotation.

Score your prompt

Score your prompt against this failure mode in 30 seconds

Paste your prompt and the platform you intend to use. AVA returns a red/yellow/green score against this specific failure mode plus a concrete rewrite if the risk is high.

Install free All 105 failure modes

AVA Pro · founders' round

$50 for 6 months of unlimited scoring across all failure modes + personal failure-history dashboard. Locks in $13/mo grandfathered after.

Claim $50 founders