B2B content with AI without losing authenticity: what Globo and IAB show

73% of marketers already use AI to create content, according to IAB Brasil and Nielsen. In our reading, the risk for a B2B brand isn’t using the tool, but publishing the text as it came out of it. What protects authenticity is your own thesis, verified data and an author who signs.

What the numbers say

The second edition of the IAB Brasil and Nielsen study (Sep/2026) shows content creation as the second most common use of AI in marketing, at 73%, behind data analysis (90%). Efficiency (80%) and speed (79%) lead the cited benefits.

The Consumidor Moderno article doesn’t state the sample, so we read the numbers as a trend.

The point for anyone who produces content is this: if almost everyone uses the same tool for speed, the average text ends up looking alike. What differentiates becomes what the tool doesn’t have.

What AI doesn’t have

It doesn’t have experience lived by your company, it doesn’t know the customer by name and it isn’t accountable for what it publishes. It also doesn’t know which data is recent and which was corrected a week ago.

That’s why we defend a process with four safeguards.

Thesis before the text. Write in one sentence what your company argues about the topic, in a way someone could disagree with. Without a thesis, AI fills in with the consensus.

Data with a source. Every number comes from whoever published it, with year and link. What AI suggests without a source doesn’t go in.

Real experience or nothing. A customer case, result or conversation only goes in if it happened and the team told it. Inventing for color is the shortest path to losing credibility.

An author who signs and reviews. The text goes out in a person’s name, and they rewrite what doesn’t sound like them.

What to review before publishing

Look for the sentences that serve any company. If the text could be on a competitor’s site without changing a word, it isn’t ready. Also cut the general-context openings and the closings that repeat what the text already said.

Signaling the use of AI

The Globo survey has a section on transparency, which we develop in AI in communication: 59% prefer, 74% want to know. The decision to disclose or not is the brand’s, and it’s worth making it beforehand, as a policy, not after someone asks.

To take to the team

1. What’s the thesis of the last text we published?
2. Does every number have a source and year in the body of the text?
3. Did whoever signs read it and rewrite what wasn’t theirs?

To build content production with your own data, get to know NorthStar’s marketing intelligence.

Frequently asked

How many marketers use AI to create content?

73%, according to IAB Brasil and Nielsen (2nd edition, Sep/2026). The most common use is data analysis, at 90%.

Does AI-made content hurt the brand?

The studies we consulted don’t measure that effect directly. In our reading, what hurts is generic content, without a source and without an author, no matter where it comes from.

How do you keep the text from looking AI-made?

By starting from your own thesis, using data with a source, including only real experience and reviewing out loud. We don’t promise to pass detectors, because they get it wrong.

Do I need to disclose that I used AI?

It’s the brand’s decision. See what the Globo survey indicates on the topic in AI in communication.

Sources

• Consumidor Moderno, on the 2nd edition of the IAB Brasil and Nielsen study, Sep/2026. consumidormoderno.com.br
• Globo Gente, The impact of AI on Brazilian consumption habits, Aug/2026. gente.globo.com