AI-generated advertising is appearing across Meta’s Facebook and Instagram platforms at a scale that raises questions about disclosure, authenticity and brand safety, according to a new report from synthetic-media security company Resemble AI. Its 2026 Synthetic Advertising Report examined 37,768 image and video files from 15,433 ads run by 509 companies across 26 industries, using the company’s deepfake detection technology.
The report found that 8.9% of the sampled ad files contained AI-generated or manipulated media. More than half of the 389 brands analyzed had run at least one AI-generated ad from their own pages. Resemble AI says the findings point to a growing verification challenge for advertisers, agencies and platforms as generative AI becomes part of creative production.
The results should be interpreted within the study’s scope: the sample covered ads on Facebook and Instagram, and the findings reflect Resemble AI’s detection methodology rather than a census of all digital advertising.
Photorealistic Creative Complicates Ad Verification
According to the report, 70% of the identified AI ads were designed to look like real photographs or video. Nearly half depicted people, often through highly realistic synthetic models. Only around one in nine identified AI ads was stylized enough to make its synthetic nature readily apparent.
The report also identified AI-generated creative designed to resemble user-generated content, including selfie-style testimonials and phone-shot videos. Resemble AI said its detection system flagged examples associated with Teladoc, MoneyLion, Dave and Tarte Cosmetics with confidence levels of at least 99%. These are findings attributed to the company’s detection engine; they do not, on their own, establish that the companies knowingly misrepresented synthetic content as genuine customer experiences.
Disclosure was rare in the sample. Only two of 3,379 AI files reviewed for disclosure mentioned AI in the ad copy, according to the report. Turkish Airlines was identified as the only brand that proactively disclosed its own ad use of AI. Resemble AI also found no C2PA content credentials among 971 files checked.
C2PA credentials provide a standardized way to attach information about digital content’s origin and editing history. Their absence does not independently prove that an asset is synthetic, but it can make provenance harder to establish.
Synthetic Audio Creates Another Detection Gap
The report’s additional analysis of nine sportsbook, prediction-market and cryptocurrency brands examined audio alongside visual content. Of 142 identified fake assets in that subset, 62 were detected only through their audio, representing 44% of the fakes in that sample.
These examples involved real footage paired with synthetic voices, highlighting a limitation of verification processes that focus exclusively on images and video. Advertisers may need to evaluate voice authenticity as well as visual manipulation when reviewing creative assets.
Resemble AI also reported that crypto and brokerage ads had the highest synthetic-media share among the industries studied, at 29.3%. The company cited OKX as an advertiser using AI in nearly half of its sampled ads.
The report separately highlighted a case involving YouTuber Elliot Choy, who reportedly recognized his apartment in an AI-generated Kalshi ad that altered his likeness. Such cases illustrate the potential for synthetic advertising to raise questions about creator consent and the unauthorized use of personal identity.
Market Landscape: Brand Safety Extends Beyond Placement
Traditional ad verification focuses on where ads appear, whether impressions are valid and whether creative content complies with platform or advertiser policies. Generative AI adds another layer: whether the people, voices, testimonials and scenes depicted in an ad are authentic, authorized and appropriately disclosed.
For agencies and advertisers, this creates practical demands for creative review, supplier oversight, consent tracking and synthetic-media detection. Platforms also face pressure to make disclosure and provenance tools more consistent.
Regulation is evolving across jurisdictions. The report points to requirements and emerging rules concerning synthetic performers and AI-generated content in New York, South Korea and the European Union. The exact obligations vary by jurisdiction and content type, so advertisers operating across markets will need to assess applicable requirements rather than rely on a single global disclosure policy.
Strategic Outlook
Resemble AI argues that brands do not need to abandon generative AI, but should establish clearer controls around its use. Its recommendations include auditing creative assets produced by agencies and partners, distinguishing synthetic backgrounds from synthetic people, disclosing AI use and monitoring for unauthorized ads using a brand’s name or a creator’s likeness.
For AdTech providers, the findings create opportunities for audio-visual detection, content provenance, creative verification and brand-protection services. However, detection confidence alone cannot resolve questions of consent, disclosure or legal responsibility.
The broader challenge is to make AI-assisted creative production compatible with transparent advertising. As synthetic media becomes more realistic, disclosure and verification may become important components of campaign governance rather than optional safeguards.
Top Insights
- AI advertising is widespread: 8.9% of sampled ad files contained AI-generated or manipulated media.
- Realism dominates: 70% of identified AI ads were designed to look photorealistic.
- Disclosure remains uncommon: Only two of 3,379 files reviewed for disclosure mentioned AI in their ad copy.
- Audio needs scrutiny: In a nine-brand subset, 44% of identified fakes were detected only through audio.
- Verification is expanding: Advertisers may need provenance, consent and synthetic-media checks alongside conventional ad verification.
Get in touch with our Adtech experts
