Table of Contents

1. Key Highlights:
2. Introduction
3. What the OM Media Trials - Zefr Study Tested and Why It Matters
4. Which AI-Generated Content Helps Brands — and Why
5. Where AI-Generated Content Harms Brands: Spam, Misinformation, and Financial Services Risk
6. The Detection Gap: Consumers Confuse Human and AI Content
7. Scale and Stakes: Why Brands Can’t Ignore AI in Media Feeds
8. Why Transparency Changes Consumer Perception
9. Technical and Policy Tools for Provenance and Verification
10. An Operational Playbook for Advertisers and Marketers
11. Creative Guidelines: When to Use AI in Brand Work
12. Ad Buying and Measurement: Moving Toward Contextual Intelligence
13. Legal, Regulatory, and Ethical Considerations
14. Platform Responsibilities and Industry Responses
15. Examples from the Field: Lessons from Real Incidents
16. Preparing for the Long Term: Governance, Skills, and Organizational Change
17. Balancing Innovation and Caution: Strategic Recommendations
18. FAQ

Key Highlights:

• A survey of nearly 5,000 U.S. and Canadian consumers finds that reactions to AI-generated video vary widely: satire, youth-focused clips, and artistic pieces can boost brand perception, while AI spam and misinformation damage it.
• Transparency matters: 41% of respondents reported improved opinions of brands when content clearly disclosed it was AI-generated; 32% already suspect human-created work is AI.
• Brands must adopt nuanced policies—mapping risk by content type and category, testing placements, demanding disclosure, and using technical provenance—to protect trust without abandoning creative opportunity.

Introduction

AI-generated video is no longer niche. It appears in snackable social clips, influencer mashups, and brand creative. Public pushback against low-quality or unsettling automated creations—lumped together as “AI slop”—has produced high-profile PR incidents and prompted some advertisers to distance themselves. Yet a blanket retreat from AI risks missing genuine opportunities. A new consumer study from OM Media Trials (Omnicom’s research arm) and brand-safety vendor Zefr reveals a more complicated picture: certain types of AI-generated video can enhance an ad’s effectiveness, while others pose clear reputational hazards. Understanding which is which, and how disclosure affects perception, gives brands a practical roadmap for using AI without eroding trust.

The findings are both a caution and a guide. They expose where advertising adjacency creates value and where it creates liability, and they point to transparency as a straightforward lever brands can use to shift consumer impressions. The next sections unpack the study’s results, explore the mechanisms behind consumer responses, and provide a playbook for marketers, media buyers, and creative teams confronting a rapidly proliferating AI content ecosystem.

What the OM Media Trials - Zefr Study Tested and Why It Matters

The study surveyed nearly 5,000 consumers across the United States and Canada to measure how people responded when ads ran after eight distinct types of AI-generated video. Rather than treating all synthetic content as uniform, researchers categorized AI video by style and intent—satire, youth-oriented depictions, artistic content, spam, and misinformation about public figures among them—and examined the downstream effect on brand perception.

Two methodological choices make the results useful for advertisers. First, the study focused on ad adjacency, the practical situation brands face when their creative appears immediately after or next to AI-produced clips on platforms and publisher sites. Second, the research paired qualitative judgment (how consumers felt about the ad and brand) with simple disclosure experiments—showing the same ad next to content labeled as AI-generated versus unlabeled—to isolate the effect of transparency.

Why this matters: programmatic buying and platform feeds increasingly serve AI-generated clips at scale. Before making broad exclusion decisions, media teams need data on which AI contexts actually harm brand equity and which can be harnessed to reinforce it. The study provides empirical foundations for those decisions.

Which AI-Generated Content Helps Brands — and Why

Consumers in the study reacted positively to ads shown adjacent to three broad categories of AI-generated video: satire, youth depictions, and artistic content. Each of these generates a context that can align with specific brand personalities and campaign objectives.

• Satire and parody: When AI is used to craft humorous, self-aware or satirical clips, audiences often treat the content as playful commentary rather than deceptive fabrication. Brands placed next to such content were perceived as “refreshing” or “innovative.” Humor creates emotional resonance; when the viewer recognizes satire, the ad benefits from the positive affect carried over from the clip.
• Youth-oriented depictions: AI-driven simulations that mirror youth culture—fast edits, meme-forward formats, stylized filters—feel native to social platforms. Ads that follow this content appear relevant and culturally tuned. Younger segments especially interpret such adjacency as authenticity, provided the creative itself respects the tone.
• Artistic content: Generative models are producing stylized, experimental pieces—visuals and animations that would be time-consuming or expensive to create manually. Audiences responding to these clips often view adjacent brands as creative or forward-looking. For categories that trade on design or innovation—fashion, entertainment, lifestyle—placement near artistic AI content can amplify the intended brand message.

Two dynamics explain these positive outcomes. First, congruence of tone: when ad creative mirrors the emotional register or stylistic cues of the AI clip, viewers perceive the experience as coherent. Second, novelty: audiences sometimes reward novelty and experimentation; well-executed AI art and satire tap the curiosity impulse rather than triggering skepticism.

Real-world illustration: When a fashion brand intentionally released AI-assisted editorial visuals that leaned into surrealism, the campaign generated buzz among design communities rather than backlash. Contrast that controlled experiment with examples where AI replaced human craft in ways that felt uncanny; the difference is intentionality and alignment with audience expectations.

Where AI-Generated Content Harms Brands: Spam, Misinformation, and Financial Services Risk

Not all synthetic content enhances an ad’s impact. The study found distinctly negative consumer reactions to ads shown after AI spam or content that presented misinformation about public figures. These contexts produce reputational risks that are difficult to inoculate against through creative excellence alone.

• AI spam: Low-quality, repetitive, or clickbait-driven AI clips provoke irritation. Viewers who encounter spammy material attribute some of that low quality to adjacent ads. The spillover effect harms brand perceptions, especially for advertisers aiming to signal reliability and taste.
• AI misinformation about public figures: The study points to a particularly acute problem: audiences are most likely to believe AI-generated misinformation about public figures. When ads appear next to content that misleads or manipulates identities, consumers associate the brand with inaccuracy or ethical ambiguity. This is especially damaging when the misinformation aligns with sensitive political or social topics.
• Financial services sensitivity: Categories that depend on credibility—banking, insurance, investment—suffer more from negative adjacency. The presence of spam or misinformation near ads for financial products erodes trust and can suppress purchase intent. When an industry’s offering hinges on reliability, even subtle trust erosion matters.

High-profile missteps illustrate the danger. Valentino’s AI-created handbag campaign drew criticism for producing “disturbing” imagery that many found unsettling rather than evocative. A McDonald’s holiday ad assembled with AI elements prompted social media backlash labeling the spot “creepy,” undermining the intended warmth of the message. In both cases the creative decision to use synthetic techniques failed to align with core brand expectations, producing more harm than value.

Programmatic environments magnify these risks because automated auctions can place ads next to millions of pieces of content with varying quality. Without fine-grained controls, a single brand buy can result in thousands of impressions adjacent to spammy or manipulative AI clips.

The Detection Gap: Consumers Confuse Human and AI Content

The study revealed a striking detection gap: 32% of respondents thought human-created content was actually generated by AI. This confusion has several implications for brands.

First, misattribution can erode earned credibility. If consumers suspect an editorial or creator-driven piece accompanying an ad is synthetic, they may downgrade the perceived authenticity of the ad itself—even when the ad is human-made.

Second, the gap amplifies the risk of false positives for brands attempting to exclude AI adjacency. If consumers already misread human content as AI, blanket blocking policies may exclude valuable, authentic inventory that audiences value.

Finally, the detection gap makes clear that creative quality and context matter more than provenance alone. If a viewer perceives a piece as fake or manipulative—regardless of its true origin—that perception shapes brand outcomes.

The proliferation of convincing deepfakes and generative models explains the confusion. Public incidents—viral deepfake videos recreating public figures, or highly produced synthetic personas on platforms—have lowered public confidence in what is real. One prominent example: a series of convincing synthetic Tom Cruise videos circulated widely on short-form platforms, fooling many viewers and sparking public debate about discernibility. Those incidents train audiences to expect deception; the side effect is a default skepticism that can damage legitimate creative.

Scale and Stakes: Why Brands Can’t Ignore AI in Media Feeds

Gartner’s projection that 90% of internet content could be generated by AI by 2030 casts the study’s findings into sharper relief. Whether or not that exact number materializes, the trajectory is toward dramatically more synthetic content in feeds, recommendation streams, and creator ecosystems.

Several practical consequences follow:

• Volume increases adjacency risk. With more AI content in inventory, brands must make careful choices about where their ads appear. Programmatic platforms and exchange ecosystems will increasingly surface synthetic clips; advertisers without controls will inadvertently fund low-quality or risky environments.
• Measurement complexity intensifies. Ad effectiveness metrics will entangle with content provenance. Existing viewability and brand lift metrics won’t capture subtle trust shifts caused by adjacent AI misinformation or spam.
• Media planning must become contextual. Historical brand-safety tools—blocklists, manual site lists—will not scale. Brands need dynamic, signal-driven approaches that evaluate content taxonomy, creative tone, and authenticity markers in near real time.

Ad buyers and creative strategists who treat AI as a niche creative tool rather than a systemic change risk being outmaneuvered. Careful governance and updated measurement frameworks must replace ad hoc responses.

Why Transparency Changes Consumer Perception

A striking outcome from the survey: 41% of respondents said their opinion of a brand improved when content was clearly labeled as AI-generated. Labeling appears to remove an element of uncertainty; disclosure allows audiences to interpret synthetic content on its own terms rather than assuming deception.

Transparency operates through two mechanisms:

• Expectation management: When a piece of content is labeled as AI-generated, viewers adjust their interpretive framework. Satire becomes clearly satirical; an artistic piece becomes an experiment in machine aesthetics. That recalibration reduces the perception of trickery.
• Attribution fairness: Disclosure signals that a brand or platform respects the viewer’s right to know. For audiences skeptical of manipulative media practices, transparency becomes a trust-building move.

A practical corollary: labeling is not a panacea. The effectiveness of disclosure depends on clarity, placement, and the underlying quality of the content. An ugly, spammy labeled AI clip will still irritate viewers. But the study shows labeling reduces the degree of brand harm in many contexts, and can even flip perception to positive in creative categories.

Technical and Policy Tools for Provenance and Verification

Beyond consumer-facing labels, several technical and industry initiatives aim to create verifiable provenance for synthetic media. Brands and publishers should monitor, pilot, and adopt these tools as they mature.

• Watermarking and content credentials: Some vendors and platforms are experimenting with invisible watermarks and metadata tags that indicate whether media was generated or edited by AI tools. Adobe’s Content Credentials is one public example of industry movement toward embedded provenance data.
• Platform labeling experiments: Major platforms have tested various forms of synthetic media disclosure—either automatically generated or creator-supplied. The efficacy of these labels depends on consistency and the platform’s ability to authenticate claims.
• Third-party verification services: Vendors such as brand-safety companies and independent auditors are developing pipelines to classify content by intent and quality. These services can offer brands programmatic signals that go beyond simple blacklists.

Limitations remain. Watermarks can be stripped, metadata can be altered, and the economics of verifying every piece of short-form video at scale are challenging. Nonetheless, embedding provenance into the content creation-to-distribution chain is one practical approach for brands seeking durable protections.

An Operational Playbook for Advertisers and Marketers

Brands must translate these insights into day-to-day ad operations. The following playbook distills the study’s implications into actionable steps media and creative teams can implement immediately.

1. Map risk by content type and product category
   • Audit your portfolio and classify campaigns by tolerance for adjacency. Financial services, healthcare, and government-facing categories will typically require stricter controls than entertainment or consumer tech.
   • Create a taxonomy of AI content types and assign risk scores—satire and artistic may be low risk; spam and public-figure misinformation high risk.
2. Use precision controls in programmatic buys
   • Move beyond site-level blocks. Demand signals at the creative-level: content taxonomy, sentiment, and provenance.
   • Work with brand-safety vendors that offer taxonomy by creative type, not just domain-level blocking.
3. Require disclosure from creator partners and platforms
   • Contractually require influencers and production partners to disclose the use of AI in campaign assets.
   • For paid placements, prioritize inventory with clear provenance or platform-level labeling.
4. Test placement and creative alignment
   • Run A/B tests that vary adjacency and disclosure. Measure brand lift, ad recall, and trust metrics rather than raw click-through alone.
   • When experimenting with AI creative, align the aesthetic and tone to anticipated audience expectations. Avoid uncanny valley effects for emotionally charged campaigns.
5. Insist on provenance and verification
   • Incorporate content-credential checks into the trafficking workflow. Where available, prefer inventory that supports machine-readable provenance metadata.
   • Use sampling and human review for high-risk buys.
6. Create crisis protocols
   • Define remediation steps if an ad appears next to manipulated or harmful AI content: immediate pause rules, public statements, and follow-up audits.
   • Monitor social sentiment and be prepared to clarify whether creative was AI-assisted.
7. Train internal teams
   • Educate brand, legal, and creative teams on the nuances of AI-generated media. Equip them with simple heuristics: assess intent, emotional tone, and perceived believability.
8. Measure beyond vanity metrics
   • Incorporate trust-related KPIs: perceived authenticity, brand favorability, and purchase intent. These signals often move in response to adjacency issues.

Implementing these steps requires investment and coordination across media buying, creative production, legal, and analytics—but the alternative is unmanaged exposure to reputational risk.

Creative Guidelines: When to Use AI in Brand Work

AI is a creative tool, and like any tool it has appropriate and inappropriate uses. The study’s findings suggest guidelines to help creative teams decide when AI will likely be an asset rather than a liability.

• Use AI for controlled stylistic experiments: Generative visuals and audio can accelerate ideation and enable novel aesthetics, particularly for youth and artistic audiences.
• Avoid unsupervised synthesis for identity or likeness: Recreating real people’s faces or voices carries high reputational and legal risk. Deepfakes of public figures ranked poorly in consumer trust and generated strong negative spillover for brands.
• Keep core brand touchpoints human-led: For hero spots—campaign-defining creative—prioritize human craft and clear authorship. Reserve AI augmentation for production-grade efficiencies: layout, color grading, or alternate cuts.
• Maintain editorial clarity: If an asset is intended as satire or art, make that clear through context and labeling. When ambiguity persists, favor clarity to avoid misinterpretation.
• Integrate audience feedback loops: Rapidly iterate based on test results and social listening before scaling AI-driven creative.

These guidelines balance efficiency gains with reputational protection. Brands that follow them can experiment with synthetic techniques while preserving trust.

Ad Buying and Measurement: Moving Toward Contextual Intelligence

Programmatic ecosystems must evolve to handle the complexity of synthetic content. Traditional brand-safety approaches—static blocklists or top-level content categories—are insufficient.

• Contextual signals should include creative metadata: tagging by tone, creator intent, AI provenance, and fact-check status. These signals allow buyers to optimize buys toward environments aligned with campaign goals.
• Dynamic whitelists and blacklists: Build lists that update automatically based on content classification. For instance, a whitelist for “artistic AI” inventory and a blacklist for “AI misinformation” can coexist.
• Integrate brand lift studies into placements: Ongoing measurement that correlates adjacency types with brand outcomes will let buyers fine-tune thresholds for risk.
• Demand transparency from supply-side platforms: Exchanges and publishers should surface content-classification signals through bid requests so buyers can pre-bid filter inventory.

Measurement will become the differentiator. Advertisers that combine real-time classification with outcome measurement will allocate spend more efficiently and avoid the worst adjacency pitfalls.

Legal, Regulatory, and Ethical Considerations

The legal and regulatory landscape around synthetic media is shifting. Governments and regulatory bodies are grappling with issues from deceptive deepfakes to political misinformation. Brands must be proactive about compliance and ethics.

• Disclosure laws: Some jurisdictions are considering or implementing rules that require clear labeling of synthetic political content and certain commercial uses of AI. Brands operating internationally should track evolving legal obligations.
• Advertising standards: Industry bodies and self-regulatory organizations are debating whether to require disclosure in paid ads that include synthetic elements. Expect pressure to adopt stricter transparency norms.
• Intellectual property and consent: Using likenesses or creative assets without proper rights may expose brands to legal claims. When AI models are trained on third-party copyrighted material, the downstream use should be assessed by legal counsel.
• Ethical stewardship: Ethical frameworks—covering consent, harm minimization, and data provenance—will increasingly inform procurement and creative briefs. Brands that articulate and enforce ethical AI standards will reduce reputational risk and align with consumer expectations.

Proactive governance—clear policies, legal review, and ethical audits—reduces downstream surprises and positions brands as responsible participants in the synthetic media ecosystem.

Platform Responsibilities and Industry Responses

Platforms and publishers have a pivotal role. Their content moderation policies, labeling initiatives, and investment in provenance tools determine the supply-side quality of AI-generated inventory.

• Labelling implementations vary by platform. Some platforms have rolled out synthetic media labels or toggles that let creators tag manipulated content. Consistency and enforcement are the hurdles; a patchwork of labels yields limited consumer benefit.
• Content moderation scale remains a challenge. Automated systems flag large volumes of content but struggle with nuance. Human review is expensive and slow, leaving a window where misleading synthetic clips can circulate.
• Industry consortia and standards: Groups focused on content provenance and media integrity are creating standards for machine-readable credentials. Broad adoption will take time but will be central to a scalable solution.

Advertisers should engage with platform partners to request better metadata, transparent labelling policies, and concrete remediation timelines when harmful AI content surfaces. Platform cooperation is not uniform, so contract language and platform selection will matter more.

Examples from the Field: Lessons from Real Incidents

Several public incidents crystallize the study’s themes.

• Valentino handbags campaign: A luxury fashion house used AI-assisted visuals in a campaign that many consumers found unsettling. The creative’s aesthetic departed from the brand’s established tone, producing negative social reaction and press scrutiny.
• McDonald’s holiday spot: A Christmas advertisement assembled with AI elements prompted critiques calling the ad “creepy,” undermining the intended emotional warmth. The placement of the ad next to user-generated AI content further complicated perception.
• Viral deepfakes: Highly convincing synthetic videos of celebrities circulated across short-form platforms, confusing many viewers. While not paid advertising, these clips shifted public expectations and increased general skepticism of visual content.

These incidents highlight consistent failure modes: mismatch between creative intent and audience expectation, inadequate disclosure, and failure to account for how adjacency changes interpretation. Brands that learn these lessons will avoid repeat mistakes.

Preparing for the Long Term: Governance, Skills, and Organizational Change

AI content management is not primarily a technology problem; it is an organizational one. The following long-term actions help firms prepare.

• Establish cross-functional governance: Create a committee including marketing, legal, media-buying, and ethics representatives to set AI usage policies and thresholds.
• Invest in skills: Media teams need to understand synthetic media taxonomy and partner with analytics to interpret brand lift data tied to adjacency. Creative teams should be fluent in when and how to use generative tools.
• Vendor due diligence: Evaluate brand-safety and provenance vendors not only on classification accuracy but on their ability to integrate signals into buying workflows.
• Scenario planning and playbooks: Develop detailed response plans for incidents—speed matters in public perception management.
• Public commitment to transparency: Consider publishing an AI usage policy for marketing and sponsorships. Public commitments build trust with audiences and provide internal guardrails.

Organizational readiness reduces reactionary decision-making and enables thoughtful experimentation.

Balancing Innovation and Caution: Strategic Recommendations

The study’s central message is practical: treat AI-generated video as a nuanced environment—one that offers both creative upside and substantial risks. The following strategic recommendations synthesize that perspective.

• Do not ban AI wholesale; categorize and decide. Blanket exclusions are blunt and may block high-value inventory. Use taxonomy-based policies.
• Prioritize disclosure. Clear, consistent labeling improves perception for many audiences. Require partners to disclose AI usage in paid placements.
• Make creative decisions audience-led. For youth and experimental markets, AI-adjacent contexts can boost relevance. For trust-dependent categories, tighten controls.
• Integrate technical provenance where possible. Favor inventory that supports provenance tagging and platform-level transparency.
• Measure outcomes, not assumptions. Use brand-lift and trust metrics to evaluate adjacency decisions and adjust rules based on observed impact.

These steps reconcile the need to protect reputation with the opportunity to innovate where AI adds genuine creative value.

FAQ

Q: Is all AI-generated content harmful to brands? A: No. The OM Media Trials and Zefr study finds that some types of AI-generated video—satire, youth-oriented formats, and artistic clips—can improve brand perception. Harm arises predominantly from low-quality spam and misinformation, and from mismatches between creative tone and brand expectations.

Q: What practical steps can brands take immediately? A: Map risk by campaign and category, require disclosure from creators and platforms, use brand-safety vendors that classify content by type, and run small-scale tests to measure brand lift before scaling.

Q: How effective is labeling AI content? A: In the study, 41% of respondents reported improved opinion of a brand when content was clearly labeled as AI-generated. Labeling reduces uncertainty and can shift perception, but it must be paired with quality control; disclosure alone does not fix poor creative.

Q: Are certain industries more vulnerable? A: Yes. Sectors that rely on credibility—financial services, healthcare, and governance-related communications—are especially sensitive to adjacency with spam or misinformation.

Q: Can technology fully solve the problem? A: Technology—watermarks, content credentials, and automated classifiers—offers important tools, but no single technical solution is sufficient. Provenance metadata and detection need consistent platform adoption and human oversight to be reliable at scale.

Q: Should brands refuse to appear next to AI-generated content? A: A blanket refusal is unnecessary and may forfeit beneficial placements. A more nuanced approach uses a taxonomy to allow safe creative adjacencies (e.g., artistic AI) while excluding high-risk categories (spam, misinformation). Decisions should be guided by measurement.

Q: How should creative teams use AI tools responsibly? A: Use AI for ideation, stylistic exploration, and efficiency—but avoid recreating real people’s likenesses without consent, and ensure the final work aligns with brand tone. Label AI involvement where appropriate.

Q: What should advertisers demand from platforms? A: Ask platforms for consistent and machine-readable provenance metadata, transparent labeling policies, and remediation timelines for harmful content. Contractual commitments to those standards should be part of premium media buys.

Q: How will this landscape evolve? A: Synthetic content will increase in volume and sophistication. Brands that develop governance frameworks, insist on transparency, and invest in contextual measurement will navigate this shift more safely and effectively.

Q: Where should brands focus their measurement efforts? A: Prioritize trust and brand lift metrics that capture perceived authenticity, favorability, and intent. Correlate those metrics with adjacency-type signals to identify patterns that inform buying rules.

AI-generated video is reshaping the media environment. The work of protecting brand equity while harnessing creative opportunity requires precise classification, transparent practices, and outcome-oriented measurement. Brands that adopt a considered, data-driven approach—rather than blanket bans or reflexive acceptance—will preserve consumer trust and find productive ways to use synthetic techniques in their advertising.