Google just shipped a feature that will add text to your video ads without asking. As of October 9, 2026, a new "Overlay on videos" asset optimization setting in Demand Gen campaigns allows Google to pull from your existing text assets and layer messaging directly onto your video creatives. The setting sits alongside "Shorter videos" and "Resized videos" in the Asset optimization panel. It is enabled by default.
For marketing leaders who spent Q3 refining creative briefs and aligning video assets with brand guidelines, this is the kind of platform change that can quietly erode control over how your message appears in market. The math question is straightforward: does automated text overlay improve conversion economics, or does it introduce creative variance that muddies attribution and brand consistency?
The Mechanics
When enabled, Google uses headlines, descriptions, and other text assets you've already uploaded to generate overlays containing key messages. These overlays appear on top of your video ads across YouTube (including Shorts, in-stream, and in-feed placements), Discover, and Gmail. The logic is sound on paper: research shows 85% of Facebook videos and 80% of LinkedIn videos are watched without sound, so reinforcing your value proposition with visible text could theoretically lift engagement.
The problem is that "theoretically" doesn't survive contact with a CFO asking why creative performance variance spiked in October.
Google's own documentation acknowledges the risk: automatically generated overlays may interfere with existing video designs, cover important visual elements, or introduce messaging that doesn't match the intended creative. If your video already has burned-in text, lower-thirds, or carefully positioned CTAs, Google's overlay could stack on top of them. The result is visual clutter that undermines the production investment you made in the first place.
The Compliance Wrinkle
There's a regulatory dimension worth flagging for anyone running campaigns in the EU, India, or New York. Google's July 2026 policy update requires that ads using image or video assets modified by AI optimization features carry disclosure labels in those jurisdictions. If Google's overlay system modifies your video, the ad will be labeled as "created or edited with AI" when served to users in those markets.
This isn't necessarily a problem, but it is a variable your legal and compliance teams should know about. The labeling requirement applies across Google Ads, Display & Video 360, Campaign Manager 360, Merchant Center, and Ads Editor. If you're running Demand Gen as part of a broader media mix and your other channels don't carry AI labels, you've introduced inconsistency in how your brand appears across surfaces. That inconsistency may or may not matter to your audience, but it will matter to your brand team.
The Control Question
Advertisers can turn off "Overlay on videos" individually or select "Disable all" to switch off all video optimization features. The path is Asset optimization in your Demand Gen campaign settings. If you haven't reviewed those settings since the feature rolled out, your campaigns may already be running with overlays enabled.
This is the pattern we've seen repeatedly with Google's asset automation features. Automatically Created Assets at the account level follow the same logic: enabled by default, buried in settings, and capable of generating headlines, descriptions, and even images that your team never approved. The operational discipline required is to audit these settings quarterly, not annually.
For Demand Gen specifically, this is the third major automation feature layered onto the campaign type in 2026. Google's August 2026 update introduced messaging from YouTube ads, travel-specific personalization, and multimodal video creation in Asset Studio. The platform is consolidating its non-search visual inventory into Demand Gen and adding AI-driven optimization at every layer. The question for marketing leaders is whether you're opting into these features deliberately or discovering them after the fact.
When Overlays Might Actually Help
There's a case for letting this run, at least as a test. 91% of businesses now use video as a marketing tool, and the competition for attention in feed environments is brutal. If your video assets were produced without text overlays and you're seeing low completion rates on sound-off placements, Google's automated overlays could provide a lift without requiring new creative production.
The right way to evaluate this is with a controlled experiment. Run a holdout group with overlays disabled, measure conversion rate and cost-per-acquisition against the overlay-enabled variant, and give it enough time to reach statistical significance. Two weeks minimum, four weeks preferred, depending on your volume. If the overlay variant wins on CAC payback, you've found a lever. If it loses, you've protected your brand from creative drift.
The wrong way to evaluate this is to let it run by default and assume Google's optimization is aligned with your business objectives. Google optimizes for its own auction dynamics. Your job is to optimize for revenue efficiency.
The Audit Checklist
Before your next pipeline review, confirm the following:
- Asset optimization settings reviewed for all active Demand Gen campaigns
- "Overlay on videos" toggled to your preferred state (enabled for testing or disabled for control)
- Legal/compliance notified if running in EU, India, or New York with AI-modified assets
- Baseline metrics captured for pre-overlay performance to enable clean comparison
This is a 15-minute task that prevents a 15-hour post-mortem when someone asks why creative performance looks different in Q4.
The Bigger Pattern
Google's direction is clear. Demand Gen has absorbed Discovery campaigns, Video Action Campaigns, and standalone Display campaigns over the past two years. It now serves across YouTube, Discover, Gmail, the Google Display Network, and Google Maps. The platform is betting that AI-driven creative optimization will outperform human-curated assets at scale.
That bet may be correct in aggregate. It is not necessarily correct for your specific brand, your specific audience, or your specific conversion economics. The only way to know is to measure, and you can't measure what you don't control.
Turn the setting off. Run the test. Model the outcome. Then decide.
