How To Run AI Ads In The US And Canada Without Getting Fined: Every State, Every Province, Every Platform

Published September 2026. AI law moves faster than liquor law, and liquor law made me rewrite a post twice. I've linked the primary source wherever one exists: the statute, the regulator, the platform's own policy page, the study itself. Read the source before you build a campaign on it.

You're going to see a small grey tag under a video this month. It'll say "AI info" or "altered or synthetic content." It'll be under an ad, from a brand you know, featuring a person who doesn't exist.

The brand didn't add it. The platform did. And if that ad ran in New York, the brand now owes the state a disclosure whether it wanted one or not.

That's the shift. For three years the AI label was a courtesy. In 2026 it became a platform rule, then a state law, then a piece of data that lives inside the file. Most brands haven't noticed because most brands haven't run an AI person in an ad yet. The ones about to will find out the expensive way.

There's a second problem underneath the legal one. The research on how people respond to AI ads came in this year, from three separate shops, and it isn't kind. I'll get to it. Law first.

If you sell anything in the US or Canada and you've been eyeing AI video because it's cheap, this is the map. The rules that apply to a brand ad, where they apply, what they cost, and what the platforms do on their own. Then the performance numbers. Then the part nobody's writing about: there's a whole category of ad that sidesteps the synthetic performer problem entirely, and it's the one that was already outperforming.

Standard warning, same as the alcohol post and the cannabis one. I'm not a lawyer. This isn't legal advice. I ran usability labs for 25 years before I ran a backyard bar, and the habit that stuck was reading the rulebook instead of the summary of the rulebook. So I read the statutes, the platform policies, and the Ottawa discussion paper, and put them in one place. Confirm it. Then run your campaign.

The three layers. Same as booze, plus one.

Three rulebooks. They stack. The strictest one wins.

Layer one is the law. In the US that's the FTC's existing deception rules plus a growing pile of state statutes, most of them narrow. In Canada there's no dedicated AI ad statute, but the Competition Act's misleading advertising rules apply, the industry code has AI guidance, and real legislation is in consultation.

Layer two is the platform. YouTube, TikTok, Meta. Each has its own definition of what needs a label, its own toggle, and its own detection system that can label you whether you toggled or not.

Layer three is provenance. New this year, and the one with no equivalent in liquor. California now requires the big AI tools to put a persistent, machine readable disclosure inside the media they generate. The label can travel with the file.

That third layer is the one to watch, because it moves the decision out of the brand's hands.

Why any of this exists: your audience asked for it

A YouGov survey of 2,557 US adults in April 2026 found that 67% had seen AI content they believed gave false, misleading, or incorrect information. A third said they couldn't reliably tell AI content from human content at all.

Meta ran its own research before it built its label. 23,000 people across 13 countries, and 82% wanted a warning label on AI content that shows people saying things they never said.

The regulators aren't inventing a problem. The viewer wanted the label first.

United States: the federal layer

There's no federal law that says "AI ad, AI label." What there is: the FTC, applying the same deception and endorsement rules it's used for decades to whatever the ad happens to be made with.

Two things to know. The FTC's Endorsement Guides were updated in 2023 and specifically cover virtual influencers, with liability spelled out for advertisers, endorsers, and the agencies in between. Your AI spokesperson is held to the same truth standard as a human one, and the brand owns what it says. The FTC's own FAQ is the better starting point than any marketing blog.

And the money. The FTC's maximum civil penalty for the rules it enforces this way sits at $53,088 per violation. That is a ceiling for specific rule violations, not an automatic fine on every AI ad, and you'll see vendor blogs turn it into "$53,088 per post." Don't. What the FTC has done is fine undisclosed endorsements, repeatedly and expensively: Google and iHeartMedia for $9.4 million, Kim Kardashian for $1.26 million, Teami for $930K. Those cases weren't about AI. They were about disclosure. The same rulebook now names virtual influencers, and no one has been fined yet for an undisclosed AI avatar in an ad. That's not a loophole. That's a queue.

Now the twist. Washington is trying to knock the state laws down. On December 11, 2025 the White House signed an executive order that created a DOJ task force with a mandate to challenge state AI laws in court, and that task force joined xAI's lawsuit against Colorado in April 2026. The FTC followed with a March policy statement arguing some state AI rules are preempted by federal consumer protection law.

What that means for you is the same thing the lawyers are telling their clients: preemption is an argument, not a ruling. Until a court says otherwise, the state law applies. Treating the lawsuit as a compliance holiday is how you become the test case.

The takeaway: federally, the AI spokesperson is an endorser and the brand owns what it says. The FTC fines disclosure failures. It hasn't fined an AI one yet. The state laws are under attack in court and fully in force in the meantime.

New York: the one that hits advertisers directly

If you read one section, read this one.

New York signed the first law in the country aimed squarely at AI people in ads on December 11, 2025. It took effect June 9, 2026. The statute, General Business Law 396-b, says anyone who, for a commercial purpose, produces or creates an ad containing a "synthetic performer" must conspicuously disclose that in the ad, where they have actual knowledge. The penalty is $1,000 for a first violation and $5,000 for each subsequent violation, enforced by the state attorney general.

What's a synthetic performer? The statute's own words, lightly trimmed: a digital asset created, reproduced, or modified by computer, using generative AI or a software algorithm, intended to create the impression of a human performer who isn't recognizable as any real, identifiable person. An AI avatar. A digital spokesperson. Simulated extras in the background of your video.

What isn't? The law is aimed at synthetic human performances, not AI use in general. AI backgrounds, image cleanup, non human graphics, a script you wrote with a chatbot, none of that triggers it. The statute exempts audio only ads, AI used solely to translate a real performer's language, and ads for movies, shows, and games where the synthetic performer is used the same way it's used in the work itself.

Three details that matter more than they look. First, the trigger is the ad appearing in New York, not where the advertiser sits, so a Canadian brand with US targeting should assume it's in scope. Second, the statute never defines "conspicuous," and as of this writing the state hasn't issued guidance, so the first enforcement actions will define it for everyone. Third, the law doesn't say what counts as one violation. Per ad, per placement, per day. Nobody knows. That uncertainty is the scariest part of a $5,000 number.

Nobody has been fined under it yet. It's three months old. You don't want to be the case that defines it.

The takeaway: an AI human in an ad running in New York needs a visible disclosure. $1,000 the first time, $5,000 every time after, and the state hasn't said how it counts. A real human in the ad doesn't trigger this law at all.

California: the disclosure inside the file

California went after the tools instead of the brands, and that's the bigger deal.

The California AI Transparency Act, SB 942 as amended by AB 853, became operative August 2, 2026, the same day the EU's AI Act transparency rules kicked in. It applies to "covered providers," meaning generative AI systems with more than a million monthly users that are publicly accessible in California. Those providers must offer a free public detection tool, let users add a visible disclosure, and embed a latent disclosure in the image, video, and audio their systems generate or alter. The latent disclosure has to be consistent with widely accepted industry standards and detectable by the provider's own tool, and it has to be as persistent as is technically feasible. The penalty is $5,000 per violation, and each day counts as a separate violation.

Read that as a creator, a brand, or an agency. The obligation is on the tool, not on you. But the biggest AI video tools clear a million users easily, so if you're using one of them, assume the clip carries a disclosure you didn't add and can't see. Anyone with the free detector can check.

Then the second shoe. From January 1, 2027, the same law puts obligations on large online platforms to detect that provenance data and give users a way to see the content's origin and modification history. How Instagram, TikTok, and YouTube implement that is up to them and the statute's scope. The direction isn't in doubt.

Two smaller California rules while we're here. The state's bot law bars using a bot to interact with someone in a commercial transaction while misleading them about it being a bot. And there's a separate stack of political deepfake disclosure laws that don't touch brand content.

The takeaway: California put the disclosure inside the file at the tool level, and from January the platforms have to surface it. The decision about whether your AI ad is identifiable as AI is increasingly not yours.

The chatbot states

A cluster of states have rules that don't touch video at all but do catch the AI assistant on your website or in your DMs.

Utah's SB 149 requires anyone using generative AI to talk to a person to disclose that it's AI when asked, and regulated professions have to say so up front. Connecticut's 2026 session produced CT SB 5, which includes chatbot controls. Oregon's HB 4154 gives people a private right to sue over an undisclosed chatbot, the sharpest enforcement mechanism on the board.

If a brand's "customer service" replies to a comment on your sponsored post and it's a bot, that's the rule you're in. And since the sale increasingly happens in a DM you'll never see, the bot in the DM is exactly where this bites.

The takeaway: if the thing replying to customers isn't a human, several states want it to say so. Cheap to comply with. Expensive to ignore in Oregon.

Colorado: the cautionary tale

Colorado passed the most sweeping state AI law in the country in 2024, then repealed it before it ever took effect and replaced it in May 2026 with a narrower law focused on automated decisions in hiring, lending, housing, and insurance. That one starts January 1, 2027. Nothing in it touches an ad.

I include it because it's the pattern. States swing big, get sued, scale back. What survives is the narrow, specific stuff: label the fake person, disclose the bot. Plan around that.

The takeaway: the broad AI laws keep getting rewritten. The narrow disclosure rules are sticking. Build for the narrow ones.

Everyone else: political deepfakes

Since 2019, most states have passed laws requiring disclosure of AI in election content. It's the most widespread AI disclosure rule in the country and it has nothing to do with selling seltzer. Unless your brand is a candidate, skip it. If your ad borrows a politician's likeness for a joke, call a lawyer before you call me.

Canada: no AI ad statute, one code, and a clock running

Canada is the opposite of the US. Instead of fifty small rules it has none written specifically for AI ads, and the document that matters most isn't a law.

Federal. There's no dedicated federal AI advertising disclosure statute. The Artificial Intelligence and Data Act died on the order paper when Parliament was prorogued in January 2025, and nothing has replaced it. That doesn't mean AI ads are unregulated. The Competition Act's misleading advertising rules apply no matter what made the ad, privacy law applies to the data, and there's a voluntary federal code of conduct for generative AI.

The industry code isn't a statute, but it matters. In October 2025, Ad Standards Canada updated its Influencer Marketing Disclosure Guidelines to cover AI. It's guidance, not law, and Ad Standards says so on the first page. But Ad Standards administers the Canadian Code of Advertising Standards, adjudicates complaints under it, and can have your ad pulled or fixed, and a complaint can roll into a Competition Act case, which does fine you. The four AI rules, as Bennett Jones summarizes them:

1. Content generated or significantly altered by AI should be disclosed, with #MadeWithAI or #AIcreated.
2. An AI influencer not being a real person doesn't remove the duty to disclose the brand relationship. The #ad still goes on.
3. An AI influencer shouldn't give a testimonial about something it can't experience. The example the guidelines use is taste.
4. Best practice is to tag a virtual spokesperson as such: #virtualinfluencer, #AIinfluencer, or #AIcreated.

Sit with rule three, because it's the whole beverage industry in one sentence. Clause 7 of the Code already says a testimonial has to reflect the genuine, current opinion of the person giving it, based on their own use of the product. An avatar hasn't tasted your soda. It can't. So it can say your soda exists, but it can't tell anyone it's good, and a disclaimer doesn't fix a claim that was never true.

Quebec. Bill 24 took effect June 12, 2026 and protects people from having their likeness deceptively reproduced by AI, timed for the October provincial election. Federally, Bill C-25 got royal assent June 18 and criminalizes political deepfakes made to mislead voters. Both are about elections, not ads, and both remind you that Quebec also wants your ad in French, which I covered in the alcohol post.

Ontario. One workplace rule that catches agencies: employers with 25 or more staff must disclose in job postings when AI screens candidates. Not an ad rule. Filed here so you don't find it later.

And now the clock. On July 23, 2026 the federal government opened a national consultation on AI transparency, and the first item on the list is detecting and identifying AI generated content. The second is helping people know when they're dealing with an AI. Submissions close September 23, 2026. It's a consultation, not a law, and the government hasn't said whether the outcome will be legislation, regulation, or a standard. But it's the first time Ottawa has asked the labeling question out loud, and the answers will shape whatever comes next.

If you're a Canadian brand with an opinion about whether your ads should carry an AI label, this is the two weeks to say so.

The takeaway: Canada has no AI ad statute yet, but the Competition Act still applies, and the industry code already says tag the AI, tag the avatar, and never let a synthetic influencer tell anyone how your product tastes. Ottawa is asking the labeling question now, and the comment window closes September 23.

The rest of the world, in one paragraph

You might not target Europe or Korea. Your ad might land there anyway. The EU AI Act's transparency obligations took effect August 2, 2026 and can reach ads that foreseeably hit EU audiences. They require a visible label on realistic synthetic content, and a platform's automatic tag doesn't count on its own. South Korea went further than anyone, requiring a label on AI generated and AI assisted ads from January 2026 and making it an offence to strip the label. My read: that's the direction of travel, and I haven't found a jurisdiction moving the other way.

The platform layer

Same as booze. The law is the ceiling, the platform is its own floor, and you clear both or you don't run. The difference here is that all three platforms also detect and label on their own, so the toggle is less a choice than a head start.

YouTube

YouTube requires creators to disclose content that is meaningfully altered or synthetically generated when it looks realistic. Three triggers: a real person appearing to say or do something they didn't, altered footage of a real event or place, or a realistic scene that never happened. You flip the "altered content" setting in YouTube Studio and a label appears in the expanded description. Content made with YouTube's own AI tools gets labeled automatically.

What doesn't need it, per YouTube's own examples: clearly unrealistic content, animation, colour and lighting filters, background blur, and AI used for production help like scripts, ideas, or captions. YouTube has said the label doesn't affect monetization or recommendations.

TikTok

TikTok has been at this the longest. Its policy requires creators to label AI generated content containing realistic images, audio, or video, it can apply the label automatically, and it reads C2PA Content Credentials to spot AI made files on upload. There's an "AI generated content" toggle, and TikTok says using it doesn't hurt distribution as long as the content follows the rules.

TikTok also defines AI content wider than the others, sweeping in real footage that's been modified beyond minor corrections. If you ran a real clip through a heavy AI edit, you're in scope here even if you're out of scope on YouTube. Of the three, I'd call it the strictest.

Meta, Facebook and Instagram

Meta runs an "AI info" label that appears when it detects industry standard AI signals or when the creator self discloses. It started as "Made with AI," got renamed after ordinary retouching kept triggering it, and now content Meta detects as merely edited by AI gets the label tucked in the post menu while content it detects as generated by AI wears it up front. Meta reserves a more prominent label for anything it judges high risk.

Two systems to remember. Strict self disclosure with real penalties applies to political and social issue ads. Everything else gets the softer automatic label. And that gap narrows when California's platform provisions land in January.

The takeaway: all three platforms label realistic AI content and two already detect it on their own. If your ad has a synthetic human in it, the tag is coming with or without you. Toggle it yourself and control the wording.

Now the numbers. What AI ads do after the click.

The law tells you what you must disclose. The data tells you what happens after you do. Three studies landed this year, and I'm going to give you the one that's good for AI first, because I'd rather you hear it from me.

The case for AI ads. In January 2026, Taboola published a field study with researchers from Columbia, Harvard, Carnegie Mellon, and TU Munich. Across 500 million impressions, AI ads and human ads had statistically equivalent click rates once you controlled for campaign and timing. Two footnotes from the same study. The AI ads that did best were the ones users didn't perceive as AI. And the single strongest trust signal in the whole dataset was a prominent, clearly visible human face.

So the best AI ad is the one that looks like a person. Hold that thought.

The case against. In May 2026, Ipsos and Syracuse University's Newhouse School took 20 real ads from 10 major brands, rebuilt each one with AI from the same brief, and tested all 40 with 3,000 US consumers. Only a quarter of viewers could tell which were AI. Didn't matter. Human made ads scored 14% higher on Ipsos's short term effectiveness measure, which is validated against sales, and 17% higher on its long term brand equity measure. The gap was smallest on plain product demos and biggest on anything emotional. Ipsos titled the paper "AI Ads Are Good Enough, And That's the Problem."

Read those two together. AI can match a human on the click. It loses on the measures that predict the sale and the brand. That's the difference between a metric and a result, which is the whole argument in four million views a month, and what you're really buying.

The gap between the boardroom and the feed. The IAB and Sonata Insights surveyed 505 Gen Z and Millennial consumers and 104 ad executives between October 2025 and January 2026. 82% of executives believed young consumers feel positive about AI ads. 45% of those consumers did. A 37 point gap, up from 32 in 2024. The share of consumers feeling negative rose 12 points in two years. 39% of Gen Z, the cohort every brief calls "AI native," feel negative about AI ads, nearly double Millennials. And when asked to describe a brand that uses AI in its ads, consumers reached for "manipulative" and "unethical" at twice the rate the executives did.

The people buying AI ads are the most optimistic group in the study. The people they're selling to are the least.

The takeaway: AI can win the click. Humans win the measures that predict the sale and the brand. And the audience you think is fine with it is the one that likes it least.

Slop: the word your customers are already using

There's a term for what the audience is reacting to, and it's not a polite one. Slop. Low effort, high volume, machine generated content with no human fingerprint. Your brand doesn't have to make it to get hit by it.

The numbers on how much people hate it, from this year:

• Gartner surveyed 1,539 US consumers and found half would rather spend money with brands that don't use generative AI in their consumer facing content and ads.
• Fractl asked the same question two years running: in 2025, 20% said heavy AI use would lower their trust in a favourite brand. In 2026 that share was 40%, and 54% among Gen Z. The share saying it would cost a brand their trust doubled in twelve months.
• DoubleVerify's 2026 media quality report found 42% of consumers across EMEA would feel negatively toward a brand whose ads showed up beside low quality, spam like AI content. You don't have to make the slop. Sitting beside it is enough.
• Back to the YouGov survey: 69% trust AI content less than human content, and 61% say they're unlikely to engage with something once they suspect it's AI.

Brands noticed. Aerie, Equinox, and Almond Breeze all ran campaigns in early 2026 explicitly calling out AI slop, and Dove pledged in 2025 never to replace people with AI in its advertising. "No AI" is now a selling point on the front of the ad, which tells you where the audience landed.

Put the law and the data side by side. Regulators are requiring disclosure on synthetic humans at the same moment the surveys show that's the content people trust least. The label doesn't just add compliance cost. It hands the viewer the one piece of information most likely to make them scroll.

The takeaway: half of surveyed consumers would rather buy from a brand that skips AI, the share who'd lose trust over it doubled in a year, and being seen beside slop is enough to cost you. The label makes the slop easier to spot. That's the point of the label.

So what needs a label? The rule of thumb.

Different laws, different platforms, different triggers. No single list covers all of them. But run your creative through this before anything else and you'll catch most of it.

Usually needs disclosure, and always needs a second look: an AI generated person presented as real. A real person made to say or do something they didn't. A realistic scene that never happened. A virtual influencer making a commercial endorsement. AI generated or heavily altered product imagery that changes what the product looks like.

Depends on the jurisdiction or platform: real footage heavily modified by AI (TikTok, Canada's "significantly altered" standard). A synthetic voice. Synthetic extras in the background (New York, yes). A chatbot replying to customers (Utah, California, Connecticut, Oregon). Anything touching an election.

Usually doesn't need an AI label just because AI was involved: a script drafted with AI. Colour grading, background blur, captions. AI generated background art behind a real person. Obviously unrealistic or animated content. Product photos in a carousel. And a real human filmed doing a real thing with a real product.

One thing "no AI label" never means: "no rules." The #ad still goes on. The claims still have to be true. The liquor and cannabis rules still apply to liquor and cannabis. The platform's ad policy still applies. This post is about one specific new category of disclosure, not a permission slip for the rest.

Notice what's in that last bucket, though. Every ad I've ever made for a brand.

The part nobody's writing about

Every law above, every platform rule, every provenance requirement, is aimed at one thing: content that could fool a viewer into thinking a fake person or a fake event is real. The regulators are drawing a box around synthetic humans.

Creator footage is outside the box. Not because of a loophole. Because there's nothing synthetic to disclose. When I stand in my kitchen and open a can, the person is real, the kitchen is real, the can is real, the reaction is real. I still owe the #ad, I still can't lie about the product, and if it's beer I still owe every rule in the alcohol post. But New York's synthetic performer law doesn't apply. California's latent disclosure isn't in the file because no AI tool made the file. Meta's detector finds nothing. Ad Standards Canada gets a testimonial from someone who has, in fact, tasted the soda.

Go back to the Taboola study. The best AI ad is the one that looks like a person and doesn't get recognized as AI. That's a strange thing to spend money on when a person costs about the same and never gets recognized as AI because he isn't. I did the real math on cheap AI video versus a human creator earlier this year, before the label laws landed. The label laws made the math worse for AI, not better.

I came up in software, and the thing usability testing beats into you is that people don't do what they say, they do what they do. What they do is watch a real person longer. My average watch time runs past the length of the video, meaning people finish and replay it. That isn't a trick. It's what happens when the viewer's brain doesn't have to spend the first two seconds asking "is this real?" I wrote about why that emotional hook is the whole game in what makes a creative ad work, and the sales numbers on creator content are in does UGC actually work for Canadian brands.

The compliance argument and the performance argument point the same direction. That doesn't happen often. It happened in Ontario's liquor rules and it's happening here.

The AI search angle, because it's the same asset

One more reason this matters past the campaign window. Your buyer's journey now runs almost entirely through Instagram, TikTok, and YouTube, and the content that keeps working after the spend stops is the content AI search engines cite when someone asks "what's a good [your product]." I've written about how ChatGPT decides which brand to recommend and how to be the AI answer with video.

I'll be careful here, because I don't have proof that a provenance tag makes an AI search engine rank a file lower. What I do know is what those systems reward: content that answers a real question, from an identifiable source, with a real experience behind it. A synthetic spokesperson can look real. It can't have used the product, and it can't be the person a search engine attributes the answer to. The footage that's clean under the law is the footage that has that evidence built in. Instagram is a search engine now. Feed it something real. And if you want the long form version that gets you named in the answer, that comes out of the same shoot as the short form that sells.

How I'd run a campaign this quarter

This one is for the marketing manager who has to ship by Friday, not the legal team.

1. Decide the human question first. Is there a person in the ad? Real or synthetic? If synthetic, you now have a New York disclosure, an Ad Standards tag, three platform toggles, a provenance file, and an Ipsos study to argue with. If real, you have #ad and the ordinary rules. Pick before you brief.
2. If you go synthetic, write the label into the creative, not the caption. New York wants it conspicuous. The EU says a platform tag alone doesn't count. Put it on screen.
3. Keep the testimonial honest. In Canada, a synthetic spokesperson shouldn't say it tastes good. In the US, the FTC treats it as an endorser and holds the brand to what it says. Claims about experience come from someone who had one.
4. Toggle the platform label yourself. They may detect it anyway. Disclosing it yourself lets you control the wording and avoids the "high risk" version of the label.
5. Check your geo. A Canadian ad with US targeting can land in New York's scope. A US ad with any European reach is in the EU's.
6. If the chatbot replies, the chatbot says it's a chatbot. Cheapest fix on this list.
7. Test, then test again. I run creative organically first, cheap, no ad spend, find the version that moves, then beat it before any paid budget. One round is a data point pretending to be a pattern. Real footage is easier to iterate because you're not paying a vendor per regeneration.
8. Read the source, not the summary. Including this one.

The one paragraph cheat sheet

Remember nothing else, remember this. The US has no federal AI ad label law, but the FTC treats an AI spokesperson as an endorser, holds the brand to what it says, and has a long record of fining undisclosed endorsements. New York requires a conspicuous disclosure on any ad with a synthetic human, $1,000 the first time and $5,000 every time after, and hasn't said how it counts. California makes the big AI tools embed a disclosure in the file and, from January, makes the platforms surface it. A handful of states want your chatbot to admit it's a chatbot. Canada has no AI ad statute, but the Competition Act still applies, Ad Standards says tag the AI, tag the avatar, and never let it give a testimonial, and Ottawa is asking the labeling question with comments closing September 23. YouTube, TikTok, and Meta all label realistic AI content and two already do it automatically. Human ads beat AI ads by 14% on the measure that predicts sales and 17% on brand, half of surveyed consumers would rather buy from a brand that skips AI, and the share who'd lose trust over it doubled in a year. And a real human, on real camera, with a real product, sidesteps the whole synthetic performer problem. You still owe the #ad.

Questions to ask any creator or vendor before you spend

Same close as always. These double as the recap.

• Is there a synthetic human anywhere in this creative? Background extras count.
• If yes, where does the disclosure sit, and is it on screen or buried in a caption?
• Which tool made the file, and does it carry a provenance disclosure we can't see?
• Can the person making claims about the product have used it?
• Which states and provinces does the targeting reach, and have we checked New York and the EU?
• Is the commercial relationship disclosed, AI or not?
• If a bot replies to comments, does it say so?
• What's the goal, and how are we measuring it beyond views and clicks?
• How many rounds of testing happen before paid spend?

A creator who answers all of those before you ask is a creator who won't cost you the account or the attorney general's attention. I answer them up front, in the brief, because I'd rather lose a deal on honesty than win one on a label I'd have to add later.

If you're hiring for that campaign, two more pieces of homework: how to vet a UGC creator portfolio before you pay, and agency vs marketplace vs hiring direct. Then go make something you don't have to label.

Let's talk.