Every enterprise buyer we talk to now opens with some version of the same question. What are you doing about GEO? Generative engine optimization. Answer engine optimization. AI visibility. The label keeps changing, but the worry underneath it is constant and correct: people are asking ChatGPT, Claude, Gemini, Perplexity, and Grok the questions they used to type into Google, and the answer that comes back is shaping reputation, demand, and trust before a brand ever gets a say.

The category ported the wrong mental model

Look across the GEO tools on the market today and you'll notice they're all running the same playbook. Pick a set of prompts. Fire them at a handful of models. Count how often your brand shows up. Score the sentiment. Rank yourself against competitors. Wrap it in a dashboard and call it share of answer.

That's a real thing worth measuring. It's also, almost exactly, the SEO rank tracker rebuilt for a new surface. The industry looked at large language models, saw something that produces a ranked-feeling output, and reached for the most familiar tool in the drawer. Track the position. Watch the position move. Try to nudge the position up.

The problem is that an AI answer isn't a search ranking. It only looks like one.

You're optimizing a reflection

The rank-tracker framing gets one thing wrong at the root. A model doesn't store a fixed opinion about your brand that you can climb up a leaderboard. It synthesizes an answer, in the moment, from the material it was trained on and the sources it retrieves. That material is the open web, the news cycle, the forums, the reviews, the analyst notes, the social conversation. In other words, the exact narrative ecosystem that has always determined how the world sees you.

What the model says about you is downstream. It's the residue of narratives that already exist out in the world. So when a GEO tool tells you ChatGPT now describes your product as expensive, or names a competitor first, it's handed you a reflection in a mirror. Useful to see. But you can't fix your face by polishing the glass.

This is why so many teams feel stuck after they buy a monitoring tool. They get a number that moves and no real lever to move it with. The standard advice that follows, write more structured content, add schema, chase citations, is fine hygiene. It isn't a strategy, because it never asks the only question that matters: which specific narratives, in which specific places, are teaching the model to say this, and what do we do about them at the source.

There's no editor to call

We saw this play out firsthand. We had a public sector client in the middle of a geopolitical crisis. We knew the news would cover it. We expected the trolling on social, because that part always comes. What we didn't expect was the engine itself becoming a source of the problem.

People were asking Grok about the situation, and it was answering with misinformation, dressed up with citations that didn't exist. The model had invented its sources. To be fair, X has put real investment into mechanisms like community notes, and that work matters. Filing them well, quickly, and with solid sourcing is one of the levers we help clients pull. But it's one lever, not control, and no single correction mechanism moves at the speed of a crisis on its own.

The reframe, for us, was this. When a newspaper gets a story wrong, you can call the editor. There's a masthead, a corrections policy, a person on the other end of the line. There's no editor of Grok to call. You can't file a correction with a model. Every answer is generated fresh, from whatever narrative supply happens to be available to it in that moment.

So the only real lever left is the supply itself. If you can't edit the answer, you change what the answer is built from. That's not a GEO trick. It's a narrative truth that just shows up at its most unforgiving on the AI surface.

And it cuts the other way too, which is the part I find most exciting. Earned media has always been the hardest thing in our world to measure. You place a story, it runs, and then you squint at impressions and sentiment and hope it landed. The real effect on what people believed was slow, indirect, and mostly a guess. Now there's a new audience reading everything, all the time, and turning it into answers. When a narrative shifts, you can watch it move into what the models say, close to real time. The same surface that makes misinformation so dangerous is the one that finally makes narrative impact observable. Earned media just got something like a live scoreboard.

Our thesis: shape the supply, not the reflection

We think GEO is a narrative problem wearing a search problem's clothes. You don't optimize an AI answer directly. You influence the supply of narrative the answer is built from. The unit of work isn't the prompt result. It's the upstream content that feeds it.

That belief isn't new for us. It's the same conviction we've held since we started tracking how information moves online and took PeakMetrics to market in 2020. We've spent years building a system that watches how narratives form, spread, and mutate across media and social and the broader web, and that traces a claim back to where it started. GEO didn't require us to invent a new company. It required us to point the system we already had at a new surface, because that surface is fed by the same rivers we've always been mapping.

What AI Perceptions does differently

This is where the methodology gets concrete, and where we part ways with the point tools.

Start with how we see the world, because the product falls out of it. We don't think in channels. We think in narratives: single stories that form somewhere, travel, and resurface in new places, and our whole job is to follow them wherever they go. That's what narrative intelligence means to us. An AI answer isn't a new discipline that needs its own tool. It's just the newest place a narrative comes to rest.

So in PeakMetrics, AI Perceptions isn't a standalone product bolted onto the side. It lives inside the same repository as everything else we monitor. The same mention object. The same enrichment. The principle we hold across the whole platform is one repository, many ways to see it, and an AI answer is just another channel in that same record. It comes in, it gets enriched, it goes out to wherever your team works. In, enrich, out. The same as a news article or an X post or a podcast mention.

Because it sits in one record, we can do the thing a standalone tracker can't. We can connect the reflection back to its source. When a model says something about you, we're not stuck reporting that it happened. We can show the provenance: the article, the thread, the review, the claim that taught it. We can show you the same narrative surfacing in the news two months ago, then in social, then in the model's answer today, as one continuous story rather than three disconnected dashboards.

That maps to how we think about the work in three moves.

Detect what the AI is saying about you, across the engines that matter, in the prompts your buyers really use.

Decipher why it's saying it, by tracing the answer back to the narratives and sources feeding it, so you're looking at causes and not just symptoms.

Defend by acting upstream, on the specific content and conversations shaping the model, with the rest of your narrative operation in the same place rather than in a separate silo. The brands that get ahead of this are the ones who stop treating owned media, earned media, and AI visibility as three separate problems and start managing them as one picture. The same story runs through all three. Seeing them together is how you steer it instead of reacting to it.

A monitoring tool stops at detect. It gives you the number. We think the number is the least interesting part. The leverage is in decipher and defend, and you can't get there without provenance, and you can't get provenance without the source data already in the same system.

Why integration is the whole point

If you remember one thing from this letter, make it this one.

A standalone GEO tool is, by design, looking at the last mile of a story it never saw the start of. It can tell you the weather. It can't tell you the climate. The moment you treat AI answers as their own isolated channel, you've cut yourself off from the only context that explains them, and you're back to polishing the mirror.

Putting AI Perceptions in the same record as the rest of the narrative ecosystem isn't a packaging convenience. It's the methodology. It's what turns a vanity metric into an actionable one. The brands that win the AI answer era won't be the ones who got cleverest at gaming prompts. They'll be the ones who understood and shaped the narrative environment the models learn from, which means treating AI visibility as one expression of a single reputation problem instead of a new department.

That's the bet. It's the same bet we've been making since long before GEO had a name, now pointed at the surface where more and more decisions are getting made.

If you're evaluating GEO and your main question is which tool counts mentions best, I think you're asking the wrong question. The right one is: when the answer about me changes, will I know why, and will I be standing in a place where I can do something about it. That's the product we built.

Happy to show you what it looks like on your own narratives.

Nick

You’re tracking the news coverage. You’re following the social conversation. But are you missing what AI is telling people about your brand?

Those answers can draw on the same coverage, conversations, and webpages your team already works to understand and influence. When a damaging claim starts spreading, seeing those sources together can make the difference between knowing there’s a problem and knowing how to address it.

That’s why we built AI Perceptions, PeakMetrics’ new Generative Engine Optimization (GEO) solution purpose-built for communications teams. PeakMetrics is the first narrative intelligence company to measure AI-generated brand perceptions alongside news, social media, and other online sources within its core offering.

The customer stories below show why that connection matters: a national theater chain discovered outdated information feeding AI answers, while a major media organization used broader narrative intelligence to uncover the context needed to counter a conspiracy theory.

A national theater chain discovers its own website is fueling a false claim in AI answers

[Insert theater-chain image]

A false, damaging narrative about a national theater chain was gaining traction across news and social media. The communications team could see the story spreading. What it needed to understand was where the claim was coming from.

That distinction mattered for the response. Identifying the accounts and outlets repeating the story could show its reach, but the team also needed to find the information allowing it to keep resurfacing.

Following the narrative back to its source

PeakMetrics detected the narrative early and by leveraging AI Perceptions, it traced it to an outdated page on the company’s own website. The content was surfacing in large language model responses and helping fuel the false claim.

The investigation revealed a direct connection between information the company published, what AI was repeating, and the narrative circulating across news and social media.

It also gave the communications team a concrete place to intervene. The company controlled the webpage contributing to the confusion.

The finding changed the response: correcting the public conversation also meant addressing the outdated information feeding it.

Turning the finding into action

PeakMetrics recommended updated copy for the webpage, along with a public response to correct the record.

Those recommendations addressed two parts of the problem together. Updating the page tackled the outdated source information. The public response addressed the claim already reaching audiences through coverage and social conversation.

For the communications team, this provided a focused course of action. It could explain the facts while addressing a source that was helping the false story persist.

What changed for the customer

The case study shows that the narrative was corrected, the spread of misinformation across news and social slowed, and reputational impact was limited. The brand was able to respond before the false story had more opportunity to define the conversation.

The practical lesson extends beyond this incident. An old webpage can continue shaping reputation long after it stops receiving internal attention. When its content appears in AI answers, that information can reach audiences in ways a communications team may not see through media monitoring alone.

This is where bringing AI responses, news, and social data together becomes valuable. The team gains a fuller view of how a claim travels and a clearer understanding of where it can act.

For this customer, the benefit was specific: identifying an owned source contributing to the problem and using that finding to guide the correction.

Why these stories matter for GEO and brand reputation

The story shows how a more precise understanding of a narrative can change a communications team’s response.

For the theater chain, the investigation identified outdated information on an owned webpage appearing in AI answers.

AI Perceptions brings that investigative approach to how ChatGPT, Claude, Gemini, Grok, and Perplexity describe a brand. Communications teams can track answers over time, measure favorability and messaging, understand what’s being asked, investigate cited sources, and examine those findings alongside news and social conversation.

That helps teams answer the questions these customer stories bring into focus:

• What information is contributing to this portrayal of our brand?
• Where else is the same narrative appearing?
• What can we update, clarify, or correct?
• How do the answers and broader conversation change after we respond?

For communications teams, this is a practical way to put GEO to work. It connects AI monitoring to the reputation decisions they already make every day.

We built AI Perceptions to help teams see those connections and act with better information. As the theater-chain story demonstrates, the most useful discovery may be a source your team already has the ability to change.

See what AI is saying about your brand and where your team can make an impact. Request a demo of AI Perceptions.

They might ask whether your company is trustworthy, how your product compares with a competitor’s, whether you’re a good place to work, or about an emerging brand story. The answer could reflect the reputation you’ve worked hard to build. It could also repeat an outdated claim, overlook an important part of your story, or frame your brand around an issue you thought was behind you.

For communications teams, that raises a practical question: how do you understand what AI is saying about your brand and decide what to do about it?

We built AI Perceptions to help communications teams answer that question, connecting AI-generated brand perceptions with the news, social conversations, and narratives they already monitor in PeakMetrics. And we’re the first narrative intelligence company to do it.

Why we built AI Perceptions into PeakMetrics

In conversations with customers, the same question kept coming up: “What is ChatGPT (or other models) saying about us?”

But answering that question once only gets you so far. If a model describes your company as untrustworthy, your team needs to know which questions produce that answer, whether other models say something similar, and what information those responses cite. You also need to understand how that portrayal relates to the coverage and conversations you’re already tracking.

We believe GEO belongs in the broader reputation conversation. The news, reviews, online communities, and emerging narratives that communications teams work to understand and influence can also inform AI-generated answers. Seeing those signals together gives teams the context to act.

That’s why AI Perceptions is part of PeakMetrics’ core platform, with AI answers measured alongside news, social media, blogs, podcasts, broadcast, and custom sources. Your team can investigate a changing narrative in real time and see whether it begins appearing in AI responses, all within the same platform.

Here’s what that makes possible.

Track how AI answers the questions that matter to your brand

A prospective customer asks whether your product is worth buying. A job candidate asks about your workplace culture. A journalist asks about your company’s history.

Each question can surface a different version of your brand.

AI Perceptions collects fresh answers daily across ChatGPT, Claude, Gemini, Grok, and Perplexity, giving your team an ongoing view of the answers those questions produce. You can start with recommended high-volume questions about your brand or category and add your own questions around the issues that matter most to your business.

That helps you focus your monitoring on meaningful questions: Are our products described accurately? Is an old controversy still defining us? Are the messages we want audiences to understand coming through?

Instead of relying on occasional manual checks, you have a consistent baseline to work from.

Measure brand favorability, trust, and messaging in AI answers

An answer can mention your company and still leave someone with the wrong impression.

AI Perceptions helps you evaluate how your brand is portrayed, including favorability, trust, key themes, and alignment with your messaging. With customizable Smart Categories, you can also assess competitive share of voice, tone, or language that could influence a purchase decision.

For example, a company investing in innovation might want to know whether AI still describes it as a legacy provider. An employer might want to understand whether answers about its culture reflect recent changes or continue to emphasize older criticism.

Your team can measure the aspects of reputation that matter to your strategy and use those findings to sharpen messaging, identify information gaps, and prioritize communications work.

Spot changes in AI brand perception and prioritize your response

When you’re tracking multiple topics across multiple models, knowing where to look first matters.

AI Perceptions highlights the questions with the biggest week-over-week shifts, along with the questions producing the most favorable and unfavorable answers. An overall AI Perception Score gives you a benchmark to track, while model comparisons help you see where answers differ.

If perception declines, you can investigate the specific question, model, and response behind the change. You can also review summaries of how answers have evolved, where to focus efforts, and open the full responses for context.

That gives your team a more focused starting point for investigation and a clearer way to explain to leadership what changed and why it deserves attention.

Investigate cited sources to guide your communications strategy

Finding an inaccurate or damaging AI answer is frustrating when you don’t know where to begin addressing it.

AI Perceptions captures the sources cited in tracked responses and shows which domains and pages appear most often. You can investigate whether answers reference your own website, third-party coverage, community discussions, or other sources.

Those citations give your team concrete leads. An outdated page on your site may need updating. A recurring misconception may reveal a gap in your public information. A publication that appears frequently in relevant answers may deserve closer attention in your earned media strategy.

Citations don’t explain every factor behind an AI response, but they provide a practical starting point for understanding the information it presents and deciding where to focus.

Benchmark your brand against competitors in AI answers

The questions that matter to your brand often include other companies, too. Who leads your category? Which provider is most trusted? What are the trade-offs between your product and a competitor’s?

AI Perceptions lets you compare how brands appear across tracked questions, including their share of voice and how they’re positioned. That helps your team spot where a competitor is associated with a strength you also want to be known for, or where your own differentiators are missing from the answer.

You can use those insights to make more informed choices about the stories you tell, the proof points you emphasize, and the questions your content should answer.

Measure how AI answers change after campaigns and earned media

You earn meaningful coverage. You publish new research. You get an important message into the conversation. What happens next?

AI Perceptions gives teams another way to assess that work. Establish a baseline before a campaign or media push, then track whether the themes, sources, and brand descriptions in AI answers change over time.

You can look for new coverage appearing in citations, key messages becoming more prominent, or outdated claims appearing less often. These changes provide useful evidence for evaluating your strategy and deciding what to do next, alongside your existing communications measurement. They show how tracked answers evolve; they do not, on their own, prove that a specific placement caused the change.

Questions about GEO and AI reputation management

What does GEO mean for communications teams?

Generative Engine Optimization (GEO) focuses on how brands and information appear in AI-generated answers. For communications teams, that includes whether answers represent a brand accurately, reflect its key messages, and reinforce or undermine trust. AI Perceptions supports this work by measuring brand portrayals and cited sources alongside the broader narrative environment in PeakMetrics.

How is AI reputation monitoring different from tracking brand mentions in AI?

Brand mention tracking shows whether a company appears in an answer. AI reputation monitoring examines how that company is described, which themes and claims appear, and how its portrayal changes over time. AI Perceptions helps teams evaluate favorability, trust-related language, messaging alignment, and competitive positioning across tracked responses.

Who is AI Perceptions designed for?

AI Perceptions is built for communications, public relations, corporate affairs, and reputation management teams, including agencies advising clients. It helps these teams investigate brand portrayals, monitor emerging issues, compare competitive narratives, and assess whether their messages are reflected in AI answers.

How can AI Perceptions help during a reputation issue or crisis?

Teams can track questions about an issue, review how different models describe it, and investigate the sources cited in those answers. Because AI responses can be analyzed alongside news and social conversations in PeakMetrics, teams can examine whether a narrative is appearing across those channels and identify information that may need clarification or correction.

Can communications teams correct inaccurate information in AI answers?

Communications teams can address inaccurate information in sources they control, seek corrections from publishers, and make accurate, well-supported information easier to find. AI Perceptions helps identify cited pages to investigate and track whether subsequent answers change. It does not directly edit AI models or guarantee that an update will change their responses.

Can AI Perceptions help inform earned media strategy?

AI Perceptions shows which publications, domains, and pages are cited in tracked answers. Communications teams can use that evidence to identify relevant sources, understand gaps in their brand’s story, and prioritize content or media opportunities. Citation patterns provide an additional input alongside audience relevance and editorial fit when planning outreach.

How can a brand measure progress in AI reputation?

A brand can establish a baseline using a consistent set of questions, then track favorability, key-message alignment, competitive positioning, cited sources, and its overall AI Perception Score. In AI Perceptions, these measures describe the responses collected for tracked questions. They help teams assess change over time without treating one answer as representative of every interaction someone might have with an AI platform.

Bring AI into the reputation work you already do

Communications teams already work on the stories, sources, and conversations that shape how organizations are understood. AI adds another place where that work matters.

We built AI Perceptions to help teams bring it into view: understand the answers, investigate the context, and make informed decisions about how to strengthen their brand’s reputation.

See what AI is saying about your brand and where your team can make an impact. Request a demo of AI Perceptions.

As consumers, journalists, investors, and other stakeholders increasingly turn to ChatGPT, Gemini, Claude, Grok, and Perplexity for information, AI-generated answers are becoming an important part of brand reputation. But those answers do not exist in isolation. They are shaped by the same news coverage, social media conversations, online communities, and emerging narratives that communications teams already work to understand and influence.

AI Perceptions brings those worlds together.

With AI Perceptions, PeakMetrics becomes the first Narrative Intelligence platform to measure AI-generated brand perceptions alongside news, social media, and other online data sources within its core offering.

Rather than treating GEO as a standalone search or visibility problem, PeakMetrics allows communications teams to understand AI perception at the narrative level. Teams can see which narratives are emerging around their brand, how those narratives are spreading across the information environment, when they begin surfacing in LLM responses, and how they are ultimately changing the way AI represents the brand.

“AI is becoming one of the first places people go to learn about a company, and what it says about your brand is being shaped by everything happening around you online,” said Nick Loui, CEO and co-founder of PeakMetrics. “If a new narrative starts taking off in the news or on social media, we want teams to be able to see when that starts showing up in AI answers too. More importantly, they need to understand what’s driving it and where they can actually do something about it. That’s the gap we built AI Perceptions to solve.”

With AI Perceptions, communications teams can:

• Track brand perception across LLMs, including favorability, key messages, trust, and competitive positioning.
• Know the top searched questions around your brand, product, competitors, or industry.
• Understand what is driving AI answers by identifying the narratives, citations, and sources influencing responses.
• Measure changes over time to see where AI perception is improving or declining across models and topics.
• Benchmark against competitors to understand where brands are gaining or losing ground.
• Turn insights into communications strategy by identifying where earned media, messaging, and other PR efforts can have the greatest impact.

This connected view allows communications teams to follow a narrative as it moves across the information ecosystem, understand when it begins influencing AI responses, and identify the sources and conversations driving that change.

For communications teams, this also creates a new way to measure impact. Organizations can benchmark AI perception before a campaign or media push and track whether new coverage, messages, and narratives ultimately change how AI responses represent the brand.

“What I like most about AI Perceptions is that it connects two areas we typically look at separately,” said Michael Brito, Global Head of Data & Intelligence at Zeno Group. “We can track the media and social narratives shaping a brand’s reputation and see whether and how those narratives are making their way into AI search engines. That gives us a much more holistic view of reputation, particularly during a crisis or emerging issue. GEO isn’t just about visibility in AI search. It’s increasingly another lens into reputation and how narratives move across the broader information ecosystem.”

About PeakMetrics

PeakMetrics is an AI native narrative intelligence company helping organizations detect, decipher, and defend against the narratives shaping reputation, risk, and influence. Combining an AI powered intelligence platform with forward deployed solutions engineering, PeakMetrics transforms data from news, social media, podcasts, and AI models into intelligence tailored to each organization's unique information environment. Customers use PeakMetrics to anticipate emerging issues, understand the forces driving narratives, measure reputation and AI visibility, and deploy custom workflows, applications, dashboards, alerts, and executive reporting that turn intelligence into action.

The campaign features Sweeney posing with strategically placed sports equipment to promote the sports prediction platform. Female athletes criticized its portrayal of women, bringing questions about sexualization and respect for women’s sports into the conversation.

But how much of the online response was criticism? And who was getting the attention: Sweeney, the ad or Novig?

PeakMetrics analyzed conversation across X, Instagram, TikTok, Reddit, Threads and Bluesky to understand what people were actually saying. The findings, featured in Business Insider, show a campaign that generated substantial visibility for Novig, a largely neutral-to-favorable response toward the brand and a conversation overwhelmingly centered on its celebrity spokesperson.

The athlete backlash raised a specific concern. According to the Associated Press, Olympic swimmer Ariarne Titmus, sprinter Amy Hunt and gymnast Gracie Kramer were among those who criticized the campaign. Their objections centered on the work women put into being taken seriously as athletes and whether the ad reinforced the sexualized portrayals they have fought against.

Those concerns matter for a brand seeking a place in sports. They also raise a question that overall conversation volume cannot answer: does a prominent wave of criticism reflect the wider response?

In PeakMetrics’ analysis, 49.7% of posts were neutral toward Novig, 32.8% were favorable and 17.5% were unfavorable. Favorable posts were nearly twice as common as unfavorable ones, even as the campaign attracted criticism.

The tone of the broader ad conversation followed a similar pattern. Neutral commentary accounted for 42.1% of posts, followed by complimentary responses at 29.9%. Questioning posts made up 12%, critical responses 9% and humor 7%.

That does not diminish the athletes’ objections or establish how female sports audiences specifically viewed the campaign. It shows that criticism was one part of a broader response that also included praise, jokes, questions and people simply discussing the ad.

Looking at who people were talking about adds another layer.

Nearly 68% of posts focused mainly on Sweeney, compared with 21% focused on the ad and just 4% focused primarily on Novig. In a separate measure of mentions, Sweeney was mentioned 67% more often than Novig in relation to the campaign.

She also occupied the largest share of both the praise and criticism categories. Sweeney accounted for 51% of the complimentary conversation and 25% of the critical conversation. The ad itself accounted for 26% of the complimentary conversation and 21% of the critical conversation.

For Novig, that creates a more complicated picture than the overall favorability numbers alone suggest. People could respond positively to Sweeney or enjoy the campaign without saying much about the company or its product.

Explicit praise for Novig or its product was comparatively rare: it accounted for about 1% of the complimentary conversation. The brand represented 20% of the critical conversation.

These percentages describe the makeup of two different groups of posts. They do not mean Novig received 20 times more criticism than praise. They do suggest that the company played a much smaller role in what people praised than in what they criticized.

Still, Sweeney’s dominance did not prevent Novig from gaining visibility.

PeakMetrics recorded a 911.1% increase in online conversation about Novig immediately following the ad’s launch. On X alone, the brand accumulated roughly 35,000 mentions during the post-launch period analyzed.

Before the campaign, Novig averaged around 900 mentions a day on X. After launch, that average rose to approximately 5,772, or about 6.4 times its previous level. Mentions peaked above 10,000 on September 10 and remained elevated in the period analyzed.

The campaign clearly created a moment for the company. What the data does not yet establish is whether that attention will translate into product interest, new customers or lasting trust.

For communications teams, this is why it helps to examine volume, favorability and the focus of conversation together. Novig’s visibility increased sharply. The overall response toward the brand leaned neutral to favorable. Yet most of the discussion centered on Sweeney, while athletes raised concerns directly relevant to the sports community the company hopes to reach.

Each finding answers a different question about the campaign’s impact.

The next question for Novig is whether people keep talking about the company once the immediate attention around Sweeney fades, and whether that conversation gives them a reason to try the product. The ad generated awareness. Sustained interest is what comes next.

Coxon announced his departure on X on September 8, criticizing Anthropic and his former employer, OpenAI, for pursuing increasingly powerful AI without acting responsibly about the risks.

In the two days that followed, catastrophic-risk language appeared in nearly half of social posts about the story. Yet just 1.4% called for people to stop using a named AI product, according to an analysis by narrative intelligence company PeakMetrics.

The findings reveal a divided response: concern about AI’s future, skepticism toward the person raising the alarm and comparatively little expressed intention to abandon the technology. For the two companies at the center of the story, the reputational shifts were also markedly different.

A warning met with skepticism

PeakMetrics analyzed coverage and conversation surrounding Coxon’s resignation from September 1–10 across news and social media.

The immediate discussion centered heavily on AI’s potential dangers. On September 9–10, 49.2% of analyzed social posts contained catastrophic-risk language, while only 0.4% mentioned AI’s benefits.

Across news and social mentions, losing control of AI appeared in 45.3% of the conversation, and extinction or existential risk appeared in 44.1%. Posts could discuss more than one theme.

But the prevalence of alarming language did not mean people accepted the warning.

On social media, 19.7% of posts attacked Coxon’s credibility and 15.7% expressed skepticism or dismissal. Together, those reactions accounted for roughly twice the share expressing fear or concern, at 17.8%.

Another 7.3% used humor or mockery. Parodies rewrote Coxon’s resignation around other jobs and companies, turning the warning into a format people could adapt for their own punchlines.

News coverage struck a different balance. Fear or concern appeared in 26.7% of classified news mentions, while attacks on Coxon’s credibility appeared in just 4.1%.

Many mentions in both channels simply relayed the news. Still, the contrast showed how differently the story was being interpreted: coverage gave more space to concern, while social reactions more frequently challenged the messenger or dismissed the claim.

One message disputing Coxon’s employment history accounted for 15.3% of English-language social mentions about the story. While PeakMetrics did not verify the underlying claim, its spread illustrated how a specific allegation about the researcher became a prominent part of the response to his warning.

Anthropic saw the larger shift

Although Coxon criticized both Anthropic and OpenAI, the story occupied a much larger share of Anthropic’s conversation.

Coxon-related discussion represented 27.9% of Anthropic mentions during September 9–10, compared with 11.8% for OpenAI, according to PeakMetrics.

The difference was also visible in broader brand favorability. In human-reviewed samples, Anthropic’s unfavorable share rose from 22.8% during September 1–8 to 52.3% during September 9–10. OpenAI’s unfavorable share rose from 38.8% to 45.1%.

The blame surrounding the resignation fell unevenly. While 65.0% of PeakMetrics’ blame classifications pointed to both companies together, 14.8% pointed to Anthropic alone. Just 1.3% singled out OpenAI.

Anthropic was the company Coxon had just left, giving it a more direct connection to the event driving the coverage. OpenAI was implicated in the warning, but the resignation occupied a smaller portion of its overall discussion.

Little talk of abandoning AI

Despite the severity of the risks being discussed, calls to stop using AI products remained uncommon.

Just 1.4% of analyzed social posts urged others to stop using a named AI tool. Another 1.2% said the author would stop or cut back personally.

Calls for a government or industry pause were more common, appearing in 6.1% of posts. Most posts—91.2%—contained no call to stop using anything.

The story also occupied relatively little space in the AI-focused subreddits PeakMetrics examined. Coxon mentions accounted for 0.51% of their conversation during September 9–10.

That figure applies to the monitored communities, rather than the entire online AI discussion. Even so, it points to a gap between the story’s prominence in the news and its presence in places where people routinely discuss AI products.

Automation amplified both sides

As the conversation grew, so did the share of participating accounts showing signs of automation.

PeakMetrics found automation signals in 25.6% of scored social accounts during September 9–10, up from 11.0% before the thread.

Among flagged posts, 45.5% amplified Coxon’s warning, while 31.5% attacked his credibility. Interestingly, the automated activity was more frequently associated with spreading the warning than challenging it.

The question is what happens next

The immediate aftermath of Coxon’s resignation produced a sharp contrast: catastrophic language spread widely, but calls to abandon AI products remained rare. Anthropic experienced a larger unfavorable shift than OpenAI, while a substantial share of social discussion focused on whether Coxon should be believed.

Whether those reactions persist is a separate question. Further tracking could show whether concern becomes sustained pressure on the companies, whether calls for intervention grow and whether more users begin expressing an intention to change their behavior.

For now, PeakMetrics’ findings show that the warning’s reach and its reception were different stories. Coxon brought AI safety back into the ongoing spotlight. The response split over the risks, the companies—and Coxon himself.

If you’re interested in seeing the full report or connecting with our team, reach out here.

Bots used to be easy to spot, with obvious spam and broken profiles. Today, they form coordinated networks that can look and feel convincingly human. They can amplify fringe narratives, impersonate real people, and flood hashtags with copy-pasta talking points. At PeakMetrics, we focus on how those bots shape narratives, not just that they exist. Enterprise bot detection software becomes far more than a security tool; it is a core part of narrative intelligence and media monitoring that helps protect brands, people, and operations.

How Social Media Bots Shape and Distort Online Narratives

To manage brand risk, it is not enough to know that bots are active. We need to understand the roles they play inside a conversation and how they change what everyone else sees. In practice, bots often fall into a few narrative roles that matter a lot for enterprise teams.

Common roles include:

• Narrative seeding, starting a conversation with a provocative claim or suspicious “leak” that looks organic.
• Narrative boosting, swarming existing posts with likes, shares, and replies to inflate their importance.
• Narrative distortion, drowning out authentic voices or hijacking hashtags so a topic looks angrier, bigger, or more polarized than it really is.

When your team sees a wave of attention around an issue, a few key questions quickly become vital:

• Did this narrative start with bots or humans?
• At what point did bots enter the conversation?
• Are they copying and pasting the same language or adapting based on human feedback?
• How has bot activity changed the tone, reach, or framing of the discussion?

Those details shape risk in very concrete ways. Bot-driven engagement can make fringe opinions look mainstream, prompt reporters to cover a “backlash,” or pressure executives and public officials into hasty public responses. For enterprises, that can mean: brand slander gaining traction, stock volatility driven by manipulated conversations, misinformation during crises, and targeted attacks on leadership or employees. Understanding where bots sit inside those stories puts your response on much firmer ground.

Core Signals for Reliable Social Media Bot Detection

Enterprise teams often ask what to look for when they suspect bot activity. Reliable detection usually blends three categories of signals: behavioral, content, and network.

Behavioral signals can be strong early clues:

• Posting at extremely high frequencies that would be hard for a human to sustain.
• Always-on activity across time zones with no natural breaks.
• Repetitive engagement behavior, like liking or sharing hundreds of posts in tight bursts.
• Odd follower-to-engagement ratios, such as tiny followings with huge volumes of activity.

Content signals focus on what is being said and how:

• Identical or near-identical copy-pasta messages across many accounts.
• Sudden spikes around the same phrase, hashtag, or talking point.
• Language patterns that feel generated or strangely generic for the topic.
• Repeated links to the same limited set of domains or low-quality sites.

Network signals look at how accounts connect and move together:

• Tight clusters of accounts that frequently share and like only one another.
• Newly created profiles that mostly amplify a handful of older “anchor” accounts.
• Cross-platform waves where the same narrative appears in quick succession on different networks.

For an enterprise, spotting a few suspect accounts is not enough. You need to move past single-user flags into campaign-level insight. That means identifying bot networks, the likely organizers behind them, and the narratives they are advancing compared to organic human behavior. Our work at PeakMetrics centers on that bridge: tying bot detection directly to narrative flows so your team can see when a story is being engineered rather than emerging naturally.

Choosing Enterprise Bot Detection Software That Actually Helps

Not all tools that claim to detect bots are designed for enterprise decision making. When we say “enterprise-grade,” we are talking about software that can operate at the scale and speed of major brands, agencies, and government teams.

Key expectations should include:

• Coverage of millions of mentions, not just small samples.
• Multi-language support to keep pace with global narratives.
• Cross-platform coverage across major social and online sources.
• Real-time alerts when emerging threats or anomalies appear.

Within that, certain capabilities matter most:

• Narrative-centric analysis that maps how a story starts, spreads, and evolves between bots and humans.
• Bot scoring that is explainable, so your team can see why an account looks automated, and adaptable, so it updates as attackers change tactics.
• Integrations with your existing media monitoring, crisis management, and security workflows instead of sitting in a silo.

One of the most important jobs of enterprise bot detection software is helping you distinguish between narratives that begin in human communities and those that are seeded or hijacked by bots. You also want to know when a conversation flips, when a mostly human thread becomes heavily bot-driven. With AI-powered narrative intelligence, we can extend that view into deepfakes, synthetic content, and cross-channel coordination that older rules-based tools typically miss.

Best Practices for Operationalizing Bot Detection in Your Team

Detection only delivers value if your team knows what to do when a bot campaign shows up on your radar. That is where process, playbooks, and alignment come in.

We usually recommend:

• Build clear playbooks, not one-off reactions. Define thresholds for when bot activity triggers executive briefings, PR responses, legal reviews, or security escalations.
• Align communications, security, and risk teams so they use shared dashboards, shared definitions of what counts as a bot campaign, and agreed protocols for when to intervene or when to avoid amplifying the issue.
• Treat bot detection as part of ongoing narrative monitoring, tracking not only accounts but also the topics, brands, people, locations, and issues that matter most to your organization.
• Use patterns you see in your enterprise bot detection software to run tabletop exercises, simulate narrative attacks, and stress-test crisis communication plans.

When everyone sees the same signals and speaks the same language about bot activity, your organization responds faster and with more confidence. Bot detection stops being a specialist task and becomes a standard lens across brand, risk, and security work.

Turning Bot Intelligence Into Stronger Brand and Risk Decisions

The real value of understanding bots is not just cleaning up fake accounts; it is improving how you make decisions under pressure. When you know which parts of a conversation are bot-driven and which are genuinely human, your options become clearer.

Sometimes the right move is a visible response from leadership. Other times, it is more effective to quietly document activity, share evidence with platform trust and safety teams, and avoid giving a manufactured narrative extra oxygen. Bot intelligence can feed into executive briefings, board-level risk reports, and ongoing brand health tracking, so you are looking at narrative dynamics and bot activity alongside sentiment and volume.

Over time, continuous monitoring through enterprise bot detection software reveals patterns that one-off investigations will miss. You start to see which communities get targeted, which narratives keep resurfacing around your organization, and where you are most vulnerable to manipulation. At PeakMetrics, our view is simple: bots and disinformation are not temporary glitches; they are permanent features of the information environment. Treating bot detection as a strategic layer of narrative intelligence is now foundational to protecting brands, people, and operations at scale.

Protect Your Enterprise From Bot-Driven Risks Today

If you are ready to cut through noisy automated traffic and focus on real customer behavior, our enterprise bot detection software is built to help you move quickly. At PeakMetrics, we work with your team to surface actionable signals so you can make confident, data-driven decisions. Talk with our experts to scope your use cases, review deployment options, and get a tailored rollout plan. To discuss your needs or request a demo, contact us.

Neutrogena’s response to the recent controversy appears to have helped contain the immediate volume of conversation. It did not, however, meaningfully improve how people viewed the company or its handling of the situation.

PeakMetrics analyzed online reaction following the beauty brand’s statement to understand whether it changed public opinion, reduced calls for a boycott or altered the trajectory of the broader conversation.

For communications leaders, it shows the power a statement can have in slowing the conversation, but also that slowing a controversy isn’t the same as resolving it.

The statement was widely viewed as inadequate

Among posts directly evaluating Neutrogena’s response, 58% characterized it as insincere or inadequate. Much of the criticism centered on a perceived lack of genuine accountability or evidence that anything would change.

Another 36% described the statement as “too little, too late,” with critics arguing that Neutrogena responded only after the company faced significant public pressure.

Just 4% viewed the response as meaningful or sincere. In other words, the statement did little to change the prevailing opinion of the company or ease dissatisfaction among those already engaged in the controversy.

Overall favorability toward the statement reflected the same pattern:

• 60% unfavorable
• 36% neutral
• 4% favorable

The rhetoric surrounding the response was also predominantly negative, with critical rhetoric accounting for 61.4% of classified posts.

Together, these findings suggest that audiences largely interpreted the statement through the context of what had already happened. By the time Neutrogena responded, many people had formed strong opinions about the controversy, the company’s role in it and what an adequate response should include.

The response still helped slow the conversation

Although the statement did not repair public sentiment, it appears to have helped contain the immediate attention surrounding the controversy.

Conversation volume declined meaningfully after the response was released. That drop was likely driven by a combination of the news cycle moving forward and the statement addressing some of the immediate demand for the company to respond.

Calls to cancel or boycott Neutrogena also fell from 40% before the statement to 32% afterward. That represents an eight-percentage-point decline, or a 20% relative decrease.

This is an important signal. A response can have an operational communications benefit even when audiences react negatively to its content. Neutrogena did not win over most critics, but its statement may have reduced the urgency driving people to post, share and actively call for action against the brand.

Still, the decline should be interpreted carefully. Conversation remains above Neutrogena’s typical daily volume before the controversy, and lower activity does not necessarily indicate that perceptions have improved. People may simply be discussing the issue less frequently.

The reputational risk has not disappeared

The immediate conversation is slowing, but the controversy may continue to follow Neutrogena in less visible ways.

Boycott messaging can persist long after the initial news cycle ends, particularly in social media comments. It may resurface when Neutrogena launches a new campaign, announces a spokesperson or partnership, responds to a related cultural conversation, or receives renewed media attention.

That creates a longer-term monitoring challenge. Looking only at total mention volume could suggest that the issue is largely over. Tracking the themes, intent and rhetoric within those mentions may reveal that the underlying criticism is still present and waiting for a new moment to reemerge.

For Neutrogena, the statement appears to have helped de-escalate the immediate conversation. But the data does not show the same improvement in public perception.

The larger lesson for communications teams is that response effectiveness cannot be measured by volume alone. A decline in conversation can tell you that attention is fading. It cannot tell you whether trust has been rebuilt, criticism has been addressed or the reputational risk has truly passed.

To learn more, connect with our team.

Panettiere represented Neutrogena for roughly a decade. Earlier this year, she alleged that the company sought to end her endorsement after she publicly discussed experiencing postpartum depression in 2015 while promoting the television series Nashville. According to Panettiere, her representative prevented the immediate termination, but Neutrogena did not renew her contract several months later.

As tributes to Panettiere spread online, clips and articles recounting that experience have resurfaced. The renewed attention has shifted the conversation around Neutrogena from largely neutral to overwhelmingly unfavorable.

PeakMetrics analyzed the resulting online conversation as of 3 p.m. ET Monday to understand how people are responding, what is driving the criticism and whether it is affecting their willingness to support the brand.

The Conversation Has Turned Sharply Unfavorable

Nearly three-quarters of the conversation—72.5%—is unfavorable toward Neutrogena, compared with 20.3% neutral and just 7.1% favorable.

Much of the negative discussion centers on Panettiere’s allegation that the brand tried to end a longtime partnership after she spoke publicly about postpartum depression. Her account is being shared not simply as an isolated disagreement between a company and a spokesperson, but as an example of how employers and brand partners have historically responded when women discuss mental health openly.

That framing has given the story broader resonance and moved it beyond the details of one endorsement contract.

For Many, the Allegation Has Changed How They View the Brand

Roughly 40% of the conversation either calls for Neutrogena to be cancelled or expresses an intention to stop purchasing from or supporting the brand.

Not every critical post amounts to a direct boycott call. But many posts say the allegation has altered their perception of the company and its values.

The response is especially pointed because Neutrogena primarily markets products to women. Users are questioning why a company with that customer base allegedly failed to support a longtime spokesperson when she spoke honestly about a condition that affects many women and families.

Overall, 68.7% of the rhetoric surrounding Neutrogena is critical.

Postpartum Depression Is at the Center of the Backlash

Among posts criticizing Neutrogena, 84.4% focus on postpartum depression or perceived mental-health discrimination.

Users frequently characterize the alleged decision as punishment for speaking candidly about a serious health condition. One shared reaction described the situation as “dystopian,” reflecting the broader concern that public figures may be expected to appear relatable and authentic—until that authenticity becomes uncomfortable for the brands associated with them.

Another 12.5% of the criticism focuses on Neutrogena’s alleged use of a morality clause or its broader employment practices. People are questioning whether discussing postpartum depression could reasonably violate such a clause and why the company reportedly chose not to renew Panettiere’s contract after nearly a decade together.

A Past Decision Viewed Differently Today

The renewed criticism also reflects how much the conversation around postpartum depression has changed since 2015. People are revisiting Panettiere’s account with a greater understanding of the condition and questioning whether she received the support she needed at the time.

That discussion is happening as people mourn Panettiere and remember her life and career. This was only one part of her story, but it has struck a chord with many who believe she deserved compassion when speaking openly about something so personal.

It’s also worth noting that Neutrogena was owned by Johnson & Johnson in 2015 and became part of Kenvue when the consumer-health business separated in 2023. The brand is under different ownership today, and expectations around mental health and brand partnerships have changed considerably. While the circumstances are very different, we saw a glimpse of Neutrogena’s current approach when it responded to Tate McRae’s recent makeup-wipe blunder with humor rather than distancing itself. That context doesn’t erase the concerns people are raising, but it does give Kenvue an opportunity to acknowledge them, clarify what has changed and show where Neutrogena stands today.

How this analysis was built

These insights were surfaced using PeakMetrics Smart Categories, which use LLM-powered classification to understand the context, themes, tone and author signals behind each post — rather than relying on keywords or Boolean searches. This captures nuances like sarcasm and distinguishes between rhetoric that's critical, complimentary or potentially violent, giving a much clearer picture of what's driving a conversation.

To see how Smart Categories can monitor your brand's exposure to a fast-moving narrative, request a demo.