TL;DR

Reputation management now includes the machines. Your buyers, your board, and the journalists who cover you are already asking AI about your brand—and AI answers with total confidence, whether what it says is true or not. 

Just ask Wolf River Electric, the Minnesota company suing Google for as much as $210 million after an AI Overview invented a lawsuit against them, while Google argues it isn’t responsible for what its own AI wrote. 

Meanwhile, new Muck Rack data shows 29% of organizations have no one who owns AI visibility, and 39% aren’t measuring it at all. 

Key Insights

• AI-referred traffic to U.S. retail sites grew 393% year-over-year in the first quarter, and by March, those visitors converted 42% better than everyone else, after converting 38% worse just a year earlier. The people asking AI about your brand aren’t early adopters. They’re your buyers.
• AI gets brands wrong in three distinct ways—stale, spliced, and invented—and each has a different fix. Research reported by PAN found that 31% of ChatGPT citations contained fabricated or misattributed information, and MIT found that AI uses 34% more confident language when hallucinating.
• The ownership gap is the real crisis: 73% of PR professionals call AI search visibility the next frontier, yet 29% say no one in their organization owns it, and 39% aren’t measuring it. In a third of companies, there’s no one whose job it is to decide brand perception.
• The correction mechanism for AI is the work communicators already do: fresh owned content, real earned coverage, and consistent third-party sources overwrite the material the models were leaning on. Recency bias, usually a burden, is your best friend in a correction scenario.
• This is reputation management—the discipline communicators have owned for a century, pointed at a new audience. If comms doesn’t claim it, it defaults to whoever bought the GEO tool.

Reputation Management Now Includes What AI Says About Your Brand

There’s a solar installation company in Minnesota called Wolf River Electric. By all accounts, they’re a good business with a great culture—until people started Googling them.

Because when they did, Google’s AI Overviews confidently told searchers that Wolf River was being sued by the Minnesota attorney general for deceptive sales practices, hidden fees, and misleading customers.

Except…

None of it was true. 

There was no lawsuit. They didn’t have deceptive sales practices or hidden fees. They weren’t misleading customers.

The AI stitched it together from sources that said no such thing—and presented it as fact, right at the top of the search results page.

One customer canceled a $150,000 solar installation contract after reading it. The company says it can trace more than $24 million in lost business to the false claims, and they’re now suing Google for as much as $210 million. 

After a judge ruled in January that Google fumbled its attempt to move the case to federal court, it’s grinding its way through Minnesota state court as we speak.

And Google’s defense? 

That it isn’t responsible for what its own AI wrote. The company is leaning on Section 230—the “we’re a platform, not a publisher” argument—and at least one prominent media law scholar thinks that argument might work: “The question in this case is, ‘Who wrote the AI Overview?’ Under current law, I think the answer, pretty clearly, is: ‘Not Google.’”

Let that sink in. The AI invented a lawsuit; the invention cost a real company tens of millions of dollars, and the maker of the AI is arguing—perhaps successfully—that no one is legally accountable for it.

Now I want you to think about that from your own perspective. Put your brand in that story. How anxious does it make you feel?

In last week’s Spin Sucks content, I told you there’s a brand-new risk category almost nobody at the leadership table is managing: what is AI saying about your company when you’re not in the room?

Not “are we showing up in AI answers?” We covered that in the visibility engineering playbook, and I warned you then that getting into the answer is only half the job.

This is the other half. When you do show up, is what it says true?

Your Buyers Are Already Asking

Let’s dispense with the “this is a future problem” objection first, because the data is unambiguous.

Adobe analyzed more than a trillion visits to U.S. retail sites and found AI-referred traffic grew 393% year-over-year in the first quarter of this year. And, by March, visitors arriving from AI tools converted 42% better than everyone else.

A year ago, AI traffic converted 38% worse than regular traffic. Now it converts best-in-channel.

In one year!

Which makes sense, if you think about it. Someone who asked their AI tool of choice what to buy, got an answer, and clicked through has already been pre-sold by the machine. The only difference between now and a year ago is that they did their research in the answer rather than on your website.

And that’s exactly the problem.

The conversation about you is happening somewhere you can’t see, built from sources you don’t control, delivered with total confidence to people who will never fact-check it. Your prospective customer. The journalist backgrounding a story. The board member running due diligence. The candidate deciding whether to accept your offer.

They ask. It answers. And you find out what it said only if something breaks—a deal that goes quiet, a weird question in an interview, or a $150,000 contract canceled over a lawsuit that never existed.

The Three Ways AI Gets You Wrong

When AI misrepresents a brand, it usually fails in one of three ways—and each one has a different fix, so we’ll go through them all.

It’s stale. The models heavily favor content published in the last 12 months, but when fresh material is scarce, they revert to the old pricing, the divested product line, or the executive who left two years ago. If your organization has changed and your public footprint hasn’t, AI will describe the company you used to be.

It’s spliced. This is what happened to Wolf River. The model pulled fragments from multiple sources—a regulatory action against some solar company, complaints about an industry practice—and assembled them into a single confident narrative about the wrong brand. Nothing it read said what it wrote. It “synthesized.”

It’s invented. And then sometimes it just makes things up. Research reported by PAN’s Lauren Hill in O’Dwyer’s found that 31% of ChatGPT citations contained fabricated or misattributed information—made-up CEO quotes, your research credited to a competitor, reports that don’t exist. 

And my favorite is that MIT research found AI uses 34% more confident language when it’s hallucinating.

(Of course it does. The less it knows, the louder it gets. We’ve all worked with that guy.)

Almost Nobody Owns This

Muck Rack’s State of PR research found that 73% of PR professionals call AI search visibility the next frontier for the profession.

And in the same study, 29% say no one in their organization owns it. Thirty-nine percent aren’t measuring it at all.

Yep.

You read that correctly. 

Nearly three-quarters of us agree that AI is where brand perception is being decided. And, in a third of organizations, it’s nobody’s job.

Ask the question within your organization, “Who owns what AI says about us?” and watch the hot potato begin. Legal assumes comms is watching it. Comms assumes it’s an SEO thing. SEO points at whoever bought the shiny new GEO tool last quarter. The GEO tool, bless its heart, produces a dashboard nobody reads.

Meanwhile, the machines keep answering.

We have a crisis plan for what happens when a reporter gets it wrong. We have social listening for what happens when a customer gets loud. But the channel your buyers now trust enough to convert 42% better than any other?

Crickets.

So let’s fix it! 

AI Reputation Governance in Four Moves

Let’s talk about the discipline you need to govern your AI reputation. As I talk you through this, notice that none of this is new work—it’s work you already know how to do, pointed at a new surface. For once, I’m not adding to your overly full plate! Yay, me!

Move 1: Baseline What’s Being Said

You can’t govern a narrative you haven’t read. So the first move is an audit: run the questions your buyers actually ask—not “tell me about [Brand],” but the shortlist questions, the comparison questions, the “is it true that…” questions—across the major models, and document what comes back.

Do it quarterly, at a minimum. Recency bias means the answer you got in January is not the answer your prospect is getting in June. Honestly, if you have the time, I would do it monthly. But quarterly is non-negotiable. 

Score what you find. Is it accurate? Current? Sourced from places you’d want? How does your narrative compare to your competitors’? 

This is narrative share of voice—one of the four AI visibility metrics we covered earlier this year—doing exactly the job it was designed to do.

The baseline isn’t busywork. It’s the difference between “I think we’re fine” and knowing where you’re exposed. (And you’ll remember from last week’s article that “here’s where we’re exposed and what the exposure costs” is a sentence your CFO understands in their bones.)

Move 2: Fix Your Source of Truth

When AI gets you wrong, the instinct is to be angry at the model. The productive move is to look at what the model read.

The models build their picture of you from your owned properties, from earned coverage, and from third-party sources—review sites, directories, Wikipedia, forums. 

If your about page is vague, your newsroom is two years stale, your executive bios contradict LinkedIn, and your last substantive piece of thought leadership predates the pandemic, you have left the machines to freelance.

You need a current, unambiguous, machine-readable set of facts about your company—who you are, what you sell, who runs it, what’s true—published on properties you control, structured so the models can parse it. 

This is the same anchor content and schema work from the visibility playbook, doing double duty.

And consistency is non-negotiable. When your owned content and your earned coverage tell the same story, the models get confident about the right things. When they conflict, the models guess—and that’s what causes a $150,000 contract cancellation. 

Move 3: Correct the Record With Infrastructure, Not Just Complaints

Even if you are right and have a case to fight, Wolf River taught us that the legal route is slow and may be a dead end. 

That case has been in court for more than a year, the false answers did their damage in the first few months, and Google is arguing—with real legal scholars agreeing—that it isn’t liable for what its AI writes at all.

Whatever the courts eventually decide, you cannot build your reputation strategy on the hope that they’ll decide it quickly or in your favor.

Yes, use the feedback mechanisms—the models and search engines have them, and documented corrections matter. But the durable fix is publishing. 

Because the models weigh fresh, credible content so heavily, a steady drumbeat of current owned content and new earned coverage doesn’t just add to your narrative—it overwrites the stale and spliced material the models were leaning on.

The correction mechanism for AI is the work you already do. Fresh anchor content. Real media relations—with the traditional outlets and the newsletter authors and podcasters the models also read. 

Third-party sources kept current.

Recency bias is usually framed as a burden. In a correction scenario, it’s your best friend.

Move 4: Give It an Owner and an Escalation Path

And finally, the move that fixes the 29% problem: put it in writing.

Someone owns the quarterly audit. Someone owns the source-of-truth pages. There’s a defined threshold for when an inaccuracy is a shrug, when it’s a correction request, and when it’s a crisis—and your crisis plan now has an AI section, because “the machines are telling people something false about us” is no longer a hypothetical. Ask Wolf River.

This doesn’t require a new hire or a reorg. It requires a decision. One meeting, one roles-and-responsibilities row, done.

The only question is who—and you can guess I have an opinion on that, too.

This Is Reputation Management

Strip away the technology, and look at what this work actually is: monitoring what’s being said about your organization, maintaining the sources of truth, building relationships with the voices that shape the narrative, correcting the record, and managing escalations when they go sideways.

We have a name for that. It’s called reputation management, and communicators have owned it for a century.

It’s just a different audience. It used to be journalists, analysts, and the people in the room. Now it includes the machines that answer for you when you’re not there—and unlike the journalists, the machines never call for comment.

In the visibility engineering playbook, I told you we handed SEO to marketing because it felt “too technical,” and we spent 15 years regretting it. AI visibility was the first test of whether we’d learned the lesson.

This is the second one. And it’s bigger, because this isn’t about being discovered—it’s about being true. 

Nobody else in your organization is trained for that. 

Legal reacts. SEO optimizes. Communications is the only function whose entire job is to ensure the story being told is accurate, credible, and coherent.

So claim it. Take the four moves into your next leadership meeting, alongside the number from your baseline audit, and say the sentence nobody else will: “Here’s what AI is saying about us, here’s what’s wrong, and here’s who’s fixing it.”

Be the person who owns the answer.

Find Out What Your System Can Defend

Before you can govern what AI says about you, you need to know whether your owned and earned foundation can back you up—because the second and third moves we just went through live or die on it.

The PESO Model® Diagnostic scores your paid, earned, shared, and owned media as a system, including the integration and measurement dimensions that determine whether your correction infrastructure actually functions. It’s free, takes about 10 minutes, and tells you exactly where the machines are most likely to get you wrong.

And if you want to build AI reputation governance as a real capability—the audit, the source-of-truth infrastructure, the earned strategy, the measurement—that’s the discipline the PESO Model® Certification teaches, worksheets and all.

Or shoot us an email! If you’d like an outside set of eyes on your baseline audit—or you’d rather we run it with you—that’s exactly the kind of advisory work we love.

The machines are already answering questions about your brand. The only decision left is whether you’re in the conversation.

© 2026 Spin Sucks. All rights reserved. The PESO Model is a registered trademark of Spin Sucks.