Is it legal to use AI for advertising?

The Short Answer: AI Ads Are Legal, Deception Is Not

Ask ten marketers whether AI-generated ads are legal and you'll get ten hedged answers — because the honest one is: it depends on a patchwork of FTC rules, state laws, EU regulations, and platform policies that don't always agree. An ad that's compliant on Meta can still violate New York law, and following every platform's disclosure rules doesn't guarantee you're following the law where the ad runs.

Here's the direct answer: using AI in advertising is legal — deceiving consumers with it is not. The FTC has made this distinction explicit. In the December 2025 Rytr decision, the Commission set aside a prior consent order and rejected the theory that an AI tool is unlawful merely because it could be misused, requiring instead a concrete connection between the AI conduct and an actual Section 5 violation. FTC Chair Andrew Ferguson put it bluntly in his dissent: treating a generative AI tool as categorically illegal "threatens to turn honest innovators into lawbreakers."

But the FTC's tolerance for the tool does not extend to the claims. Under its current enforcement posture, the agency continues to pursue companies that make false or unsubstantiated statements about AI capabilities. Operation AI Comply has produced more than a dozen actions since 2024, including a $48.6 million settlement with Growth Cave over claims that its AI would automate "nearly 100%" of building an online course — when users did most of the work manually.

So the real compliance question isn't "which AI tool should I use?" It's "what am I claiming, and what am I disclosing?" That risk lives in a few specific places:

• Capability claims: every AI performance claim needs "competent and reliable evidence" behind it, per longstanding FTC substantiation standards.
• Synthetic performers: New York's law, effective June 2025, requires conspicuous disclosure of AI-generated people in ads shown to NY audiences.
• EU audiences: AI Act Article 50 requires AI-generated images, audio, and video to be clearly labeled as artificial — with penalties up to €15 million or 3% of global turnover.
• Platform rules: Meta, Google, TikTok, YouTube, and X each set their own disclosure triggers and labels, and none of them substitute for legal compliance.

This is the same trust problem SkillzBench was built to address in a different corner of the AI ecosystem: nobody can tell which AI tools and skills are legitimate and which carry hidden risk, so independent verification beats marketing hype. In advertising, the same logic applies — the tool itself won't get you sued, but an unverified claim or a missing disclosure will. As Dynamis LLP's Brooke Watson frames it: "If AI changes reality, disclose it. If it only refines what's already real, document it." The law is already here; the only question is whether your ads respect it.

The Rules Already on the Books: Disclosure Laws You Must Follow Now

The most common compliance mistake advertisers make is assuming AI disclosure rules are still on the horizon. They're not. As legal analysis from Dynamis LLP puts it, disclosure laws are already here — and they carry real teeth.

New York enacted the first U.S. law requiring "conspicuous" disclosure of AI "synthetic performers" in advertising, effective June 9, 2025. According to ArentFox Schiff's advertising compliance review, the law applies to advertisers distributing visual or audiovisual ads to New York audiences — including online and social media — with escalating civil penalties. It's unique in targeting what appears in the creative output itself, unlike California, Colorado, and Utah laws focused on AI interaction transparency.

California's AB 1836 mandates disclosure and consent when AI recreates a person's image or voice for commercial use, while AB 2655 requires large online platforms to label deceptive AI content. The UK's Advertising Standards Agency issued guidance in May 2025 requiring AI disclosure whenever it could mislead consumers about authenticity or performance.

The steepest stakes sit in Europe. EU AI Act Article 50 requires AI-generated or substantially manipulated images, audio, or video to be clearly identifiable as artificial — via visible labels or metadata — and the duty applies to deployers, meaning brands and agencies, not just AI providers. Penalties reach €15 million or 3% of global annual turnover, whichever is higher.

Then there's the platform patchwork. Meta, Google, TikTok, YouTube, and X each maintain different disclosure triggers, labels, and enforcement mechanisms. As Everything PR's platform-by-platform breakdown notes, an advertiser compliant on one platform can still be out of compliance with another — or with the law where the ad actually runs. Platform compliance doesn't equal legal compliance.

A practical rule cuts through the complexity, one echoed by legal counsel Brooke Watson: "If AI changes reality, disclose it; if it only refines what's already real, document it." That means:

• Disclose conspicuously when AI creates synthetic humans, voices, or materially altered scenes — never buried in fine print.
• Document internal AI usage when it only refines real footage, so you can prove what happened if questioned.
• Track disclosure rules per platform and per jurisdiction, with periodic counsel review, because the rules keep changing.

The consumer context sharpens the case for getting this right. 72% of consumers say AI makes it hard to judge what's authentic, and 61% already assume AI appears in ads but can't identify where. Teams building AI-assisted marketing workflows — whether that's creative tooling or agent-driven pipelines of the kind SkillzBench benchmarks — should treat disclosure discipline as part of the workflow itself, not an afterthought bolted on at sign-off.

What Gets Advertisers Fined: Enforcement Cases and the Price of Inflated AI Claims

When the FTC announces a settlement, the numbers tell the story. In early 2026, the agency's Operation AI Comply sweep produced a $48.6 million settlement against Growth Cave over claims that its AI software would automate "nearly 100%" of building an online course — when users actually completed most tasks manually. That single case shows how expensive inflated AI hype has become.

The sweep, launched in 2024, has generated more than a dozen enforcement actions, and the dollar amounts keep climbing:

• FBA Machine/Passive Scaling: a $15.7 million judgment for false income guarantees tied to AI software, with alleged consumer fraud exceeding $15 million.
• Air AI: an $18 million judgment for deceptive earnings claims about its AI product.
• Active Listening: settlements of $880,000 and $25,000 against two companies over false claims that an AI-powered ad service could target ads based on smart-device conversations.
• Workado LLC: enforcement for misrepresenting the accuracy of its AI Content Detector.

The FTC's message, as legal analysts put it, is clear: marketing hype about AI gets measured against traditional substantiation requirements. "Competent and reliable evidence" must back every capability claim before it ships.

Beyond AI-specific sweeps, the FTC's Consumer Review Rule carries civil penalties of up to $53,088 per occurrence, and the agency sent warning letters to ten companies in December 2025. Private litigation adds another exposure channel: 2025 influencer class actions targeting brands like Celsius, Shein, and Alo Yoga sought damages ranging from $25 million to over $500 million for undisclosed paid endorsements.

Advertisers also face a moving target on the policy side. Executive Order 14179 has pushed federal agencies toward a pro-innovation posture, and the FTC's Rytr decision rejected liability for AI tools merely because they could be misused. Yet the FTC simultaneously proposed a policy statement on AI accuracy — asserting that distorting AI outputs can violate Section 5 — with public comment open until July 31, 2026. Meanwhile, EU AI Act transparency penalties reach €15 million or 3% of global turnover.

That shifting landscape is exactly why counsel review matters: the rules keep changing, and what passed legal review last quarter may not hold this quarter. At SkillzBench, we see the same principle apply to AI agent skills — claims about what a tool does should be verified against what it actually delivers, which is why our benchmarks measure real Lift rather than marketing promises. The price of unverified claims, as Growth Cave learned, is measured in eight figures.

Your Compliance Playbook: Substantiate, Disclose, Document

Compliance in AI advertising doesn't require a legal department — it requires a repeatable process. The companies paying $48.6 million in FTC settlements didn't set out to break the law; they published AI claims they couldn't back up. The fix starts before anything goes live.

First: substantiate every AI capability claim. The FTC measures marketing hype against traditional substantiation standards, requiring "competent and reliable evidence" for what your ads say an AI tool can do. Operation AI Comply has produced more than a dozen enforcement actions since 2024, including an $18 million judgment against Air AI and a $15.7 million judgment against FBA Machine for false income guarantees tied to AI software. The Growth Cave settlement — $48.6 million — hinged on a claim that AI would automate "nearly 100%" of building an online course when users did most tasks manually. If you can't prove it, don't print it.

Second: build a disclosure matrix by platform and jurisdiction. There is no unified rulebook. New York's synthetic performer law, effective June 9, 2025, requires conspicuous disclosure of AI-generated human likenesses in ads, while the EU AI Act Article 50 imposes its own labeling duties — with penalties up to €15 million or 3% of global turnover. Layer on the fact that Meta, Google, TikTok, YouTube and X each apply different disclosure triggers and labels, and platform compliance doesn't guarantee legal compliance. Map every campaign against both the platform's policy and the law where the ad runs.

Third: make disclosures impossible to miss. Regulators increasingly require disclosures that are clear, conspicuous, and sufficient to change consumer expectations — never buried in terms of service. Practical mechanics include:

• Embedding C2PA provenance metadata in AI-generated images and video
• Recording AI usage in IPTC metadata for every asset
• Adding AI-content labels directly to social posts
• Requiring a disclosure sign-off step in every campaign approval workflow

A useful operating rule, as Dynamis LLP advises: "If AI changes reality, disclose it. If it only refines what's already real, document it."

Finally, treat transparency as a brand asset, not a legal tax. In a Yahoo and Publicis survey, 72% of consumers said AI makes it difficult to judge what content is authentic — and consumers reject deception, not AI itself. Aerie's public pledge not to use AI-generated bodies became its most-liked Instagram post in a year. Honesty converts.

The same rigor applies to the AI tools behind your ads. That's the premise behind SkillzBench: verify the AI skills and agents you deploy — version pinning, security scanning, and trust scores — before they ever touch your ad workflow, because an unverified agent making claims on your behalf is a compliance risk too.