MAJORFailure Reference

Google Veo Watermark Bleed — Pre-Generation Risk Reference

Technical Classification

Training-Data Watermark Leak

Veo 2 and Veo 3 occasionally regurgitate watermark patterns from their training set when the generation drifts toward archival or stock-footage neighborhoods in latent space. The most common artifacts are faint broadcast-network logos in the corner, stock-agency name strips along the bottom third, and timecode bleed at the top of the frame. None of these were in the user's prompt and all of them prevent commercial use of the output.

How to identify this failure

✕Faint broadcast-network logo in a frame corner
✕Stock-agency name strip along the bottom edge
✕Translucent timecode bleed at the top of the frame
✕Wattermark intensity changing through the clip
✕Letter-shaped marks resembling a watermark logo

Real generation examples

Prompt used

"Wide establishing shot of New York at golden hour"

Failure observed @ 0:00

Translucent "GETTY" mark in lower-right corner at 0:00–0:08; timecode strip visible at top

Prompt used

"Slow pan across a beach at sunrise"

Failure observed @ 0:03

Faint broadcast-network logo bled into upper-left at 0:03–0:06

Documentation strength

If you need to escalate

VERY HIGH — training-data bleed in commercial output is a deliverability defect; Google support refunds with a documented Generation ID.

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 Training-Data Watermark Leak 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 Training-Data Watermark Leak 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 "Training-Data Watermark Leak". 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 Google Veo refund credits for watermark bleed?

Yes. Submit the Generation ID and a screenshot of the leaked mark to Google AI Studio support; the failure mode is recognised and routinely refunded.

Why does Veo produce stock-footage watermarks?

Veo's training corpus includes large amounts of broadcast and stock material. The diffusion model memorises high-frequency patterns — including watermarks — and reproduces them when the latent drifts toward those training neighborhoods.

How can I avoid watermark bleed in Veo generations?

Use specific, unusual style anchors in your prompt to push the latent away from generic stock-footage neighborhoods ("low-key noir", "anamorphic vintage glass", "Super-8 grain"). AVA flags high-bleed-risk prompts before submission.

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