nCircle
Talk Nano to us
Menu

Blog

How Nano AI checks every creator video before it goes live

A campaign with 1,000 creators puts 1,000 new videos in front of your brand in a week. Here is how Nano AI reviews each one against 11 brand-safety checks, and where people step in.

By nCircle. Sources checked 22 November 2026.

A broadcast control room with a wall of screens and operators reviewing live video.

TL;DR

  • Vetting a creator’s past posts tells you about the creator. The brand-safety risk in a campaign sits in the new video each creator films for you.
  • TikTok and YouTube both restrict political, alcohol, gambling, drug and weapons content in branded posts, and the FTC expects advertisers to train and monitor the creators they pay.
  • Nano AI runs each campaign through six agents. One of them is our own brand-safety AI, built from five tools, which checks every draft against 11 checks and returns a flag or no flag.
  • A person reviews every flagged draft and at least 3% of the drafts that pass.
  • A draft posts only after it clears, so a problem is fixed with a reshoot before anyone sees it.

When 1,000 creators make videos for one launch, a brand gets 1,000 new pieces of content in a week. Each one is filmed in a different room, with a different soundtrack and a different background. Vetting the creators before the campaign checks their history. It cannot check a video that has not been filmed yet. That is why Nano AI reviews the draft itself, before it goes live.

Brand-safety risk sits in the draft

The platforms set the floor. TikTok’s branded content policy bars branded posts for weapons, drugs and adult products, and lists alcohol and gambling as restricted categories. YouTube’s branded content policies prohibit branded content for recreational drugs and weapons, and restrict alcohol, gambling and election or political content.

The regulator sets the duty. The FTC’s guidance to advertisers says that “your company needs to have reasonable programs in place to train and monitor the influencers you pay and direct” (FTC, Endorsement Guides: what people are asking).

Put those together and the check has to run on every video, because every video is new. A creator with a spotless history can still film in front of a party poster, play a song with lyrics you cannot run, or leave a competitor’s bottle on the table. At 1,000 creators that means 1,000 reviews, and that volume only works with a machine doing the first pass.

The 11 checks every draft goes through

Every draft gets the same 11 checks. Ten of them apply to every campaign. The eleventh comes from your brief.

# Check What it looks for
1 Nudity and sexual content Explicit or suggestive material
2 Violence Weapons, gore, fights and dangerous stunts
3 Abuse Slurs and serious profanity. Mild swearing is allowed
4 Hate Hateful or sectarian content
5 Religion Content that treats a faith insensitively
6 Politics Party flags, slogans and campaigning
7 Drugs, smoking and alcohol Use or promotion in the video
8 Gambling Betting, gambling and scams
9 Private information Phone numbers, addresses and other personal details on screen
10 Competitors Another brand’s product or logo in frame
11 Claims Anything your brief does not allow the creator to say

Each check returns a flag or no flag. There is no score and no traffic light, because a reviewer has to decide one thing: does this draft need a fix before it posts? The same draft also goes through the checks against your brief, such as whether the product is shown and the disclosure is there. You can see how a review reads on our draft review page.

Nano AI reviews each draft in four steps

Nano AI is the system that runs an nCircle campaign. It is six agents, each with one job: reading the brief, building the plan, recruiting creators, handling their replies and contracts, checking every video, and counting the views.

The checking agent is our own brand-safety AI. It is built from five tools that work together: they look at the picture frame by frame, listen to the soundtrack, read the words spoken and shown on screen, and judge each one against the 11 checks. When a draft trips a check, the review follows four steps.

  1. Flag the moment. The flag names the check and the exact time in the video. For example: a political party’s flag in the background at 0:17.
  2. Write the fix. The creator gets the correction in plain words, such as “reshoot this scene without the flag in frame”.
  3. Check the new cut. The reshot draft goes through all 11 checks again, from the start.
  4. Clear it to post. Only a draft with no open flag can go live.

The creator makes the fix before posting, so the brand never has to ask for a live post to come down.

Where people step in

Nano AI does the first pass on every draft. A person reviews every flagged draft, and at least 3% of the drafts that pass with no flag. The flagged drafts are where judgement is needed. The sample of clean drafts is how we catch what the checks miss and keep the system honest.

Here is the arithmetic for an example campaign of 2,000 drafts where 100 are flagged. People review the 100 flags plus at least 57 clean drafts, 3% of the remaining 1,900. That is about 157 human reviews across 2,000 drafts. Those figures are our worked example, and the ratio is why brand safety can scale to thousands of creators without a team the size of the campaign.

Five questions to ask any partner about brand safety

Before you sign off on a creator campaign, ask these five questions. They work for any partner, including us.

  1. Do you check the draft, or only the creator? A creator audit is useful. It does not see the new video.
  2. Which categories do you check? The list should cover at least the categories TikTok and YouTube restrict in branded content.
  3. What happens when a draft is flagged? Someone should write the fix, and the new cut should be checked again in full.
  4. How much of the cleared work does a person see? Ask for the share, and whether flagged work always reaches a person.
  5. Can a video go live before it clears? The answer should be no.

FAQ

Are the checks the same for every brand?

Ten checks run on every video in every campaign. The eleventh, claims, is set from your brief, so it checks for what your category and your legal team do not allow the creator to say.

Does the review slow a campaign down?

The checks run as each draft arrives, so review keeps pace with posting. A flagged draft goes back to the creator with a written fix as soon as the flag is raised.

What does the brand see?

Every flag, the moment it was found, the fix that was sent and the re-check. Your campaign report shows how many drafts were flagged and how each was resolved.

Brand safety at the scale of thousands

A small influencer campaign can be reviewed by hand. A campaign with hundreds or thousands of creators cannot, and that is the campaign nCircle runs. Nano AI checks every draft against 11 brand-safety checks and your brief, people review every flag and a share of the rest, and nothing posts until it clears.

Send us a brief and we will show you how the checks apply to your category.