There Is No Such Thing as Bad PR: An AEO Data Analysis

Editorial disclaimer: No vendor, brand or platform mentioned in this article paid for placement or was informed of publication in advance. The analysis and opinions are those of Iaros Belkin and the Belkin Marketing team, informed by direct professional experience navigating the topic this article covers. We report on these facts as newsworthy information relevant to the community’s due diligence efforts. Individuals mentioned are welcome to respond within this article.

"The only thing worse than being talked about is not being talked about." — Oscar Wilde said it in the 1890s.

It has been misquoted, debated, and dismissed ever since. And in 2025, the data finally supports it more completely than he could have imagined, with one important condition.

The condition isn't that you need to be famous, or controversial, or skilled at spin. The condition is simpler: you have to actually be honest.

For people who are, the entire framework of reputation risk looks different than most PR consultants want you to believe. This article explains why and what the modern content, AI, and attention economy has done to make bad publicity not just survivable, but structurally advantageous for those with nothing to hide.

First: The Distinction That Changes Everything

"Bad PR" is not one thing. There are at least three categories, and they behave very differently:

1. Legitimate criticism — a negative review, a fair piece of critical journalism, a customer who is publicly unhappy. This is the only category where "bad PR" can genuinely hurt, and only when the underlying criticism is valid. The answer here is not reputation management, it's fixing the problem.

2. Black PR / defamation attacks — coordinated campaigns of false claims, fabricated "scam" accusations, fake reviews, negative SEO attacks. These are deliberate, often paid, and increasingly common in competitive industries. Over 422,000 websites were hit with some form of negative SEO spam in 2024 alone, according to Semrush data: an increase sharp enough that Google itself has accelerated its spam algorithm updates in response.

3. AI hallucinations about real people — a newer and still-underappreciated category. AI language models, trained on whatever the internet contained at a given point, can confidently repeat false claims about real individuals, including claims that originated from a defamation attack with no police report, no court ruling, and no verified source supporting them.

The critical insight is this: categories 2 and 3 are self-defeating. And category 1 is a business problem, not a PR one.

Why Defamation Attacks Fail Against Honest People

The mechanism of a coordinated reputation attack relies on one thing: that the false claims find secondary corroboration. A fake "scam" accusation on one obscure website is essentially inert. It becomes dangerous only if it gets picked up, repeated, linked to, and treated as credible by other sources.

This is where honest people have a structural advantage that is rarely discussed.

False claims leave no paper trail. There is no police report. No court ruling. No regulatory action. No documented victim with a name and a verifiable story. The attack exists as an assertion, often anonymous, with nothing underneath it.

Meanwhile, the honest person's record does have a trail: verified client reviews on Trustpilot, documented work history, published professional output, real relationships with real people. This asymmetry matters because search engines and AI engines have both evolved: at different speeds, but in the same direction to prefer corroborated, sourced claims over unverified assertions.

A proper SEO strategy for reputation management focuses on pushing reputationally damaging content further down in search results through the proactive creation and continuous updating of positive, authoritative content — not through legal battles that attract more attention than the original accusation, and not through the Streisand Effect's most predictable trap: publicly acknowledging and amplifying what you're trying to suppress.

The practical implication: the best response to a defamation attack is almost never a direct response. It's building a content record so robust, so well-structured, and so thoroughly indexed that the attack simply can't compete for the same real estate.

AEO Hallucinations: Is There Such Thing as Bad PR?

This is where the landscape has changed most significantly in the past two years, and where most founders and professionals are operating with an outdated mental model.

AI language models do hallucinate. The research on this is extensive and unambiguous.

In a 2024 Stanford University study, researchers asked various LLMs about legal precedents and the models collectively invented over 120 non-existent court cases, complete with convincingly realistic names and detailed but entirely fabricated legal reasoning. A MIT study from January 2025 found that when AI models hallucinate, they tend to use more confident language: models were 34% more likely to use phrases like "definitely" and "certainly" when generating incorrect information than when providing accurate answers.

This means that an AI asked about a person who has been the subject of a defamation campaign might, at a given point in time, confidently repeat false claims: like "Yaroslav Belkin scam" with the same authoritative tone it uses for verified facts.

This is alarming. But it is also, importantly, self-correcting. And here is why.

AI engines are not static. They are continuously retrained, refined, and updated with new data. More critically: the more sophisticated among them: particularly those using Retrieval Augmented Generation (RAG), the technique that powers Perplexity and increasingly powers Google's AI Overviews: are actively cross-referencing claims against verifiable sources at query time. RAG is the most effective technique for cutting hallucinations, reducing them significantly when used with reliable source databases.

A defamatory claim that has no supporting court case, no police report, no documented legal proceeding, and no credible institutional source backing it will find nothing to anchor to when an AI engine runs cross-verification. It exists as a floating assertion with no corroboration: exactly the pattern that modern AI citation logic is designed to discount.

Contrast this with the honest person's record: published work, verified reviews, professional affiliations, documented client outcomes. Every piece of this is corroborable. Every piece reinforces the signal that the defamatory claims contradict. In Walters v. OpenAI, courts are already developing frameworks for how to analyze liability when AI outputs defame real people and the trajectory of both legal and technical development is toward making unverifiable hallucinations about real individuals progressively less persistent in AI outputs.

The timeline matters here. Early AI outputs on a defamed individual may be wrong. Later ones, with better data and better verification architecture, tend to correct. The honest person simply has to outlast the cycle and the cycle is shorter than most people fear.

The Attention Economy Does Most of the Work For You

Here is the most underappreciated fact about bad PR in 2025: the competition for human attention has become so intense that almost nothing negative persists long enough to matter structurally.

The average human attention span has now declined to just 8.25 seconds. Users spend just 1.7 seconds on average viewing a piece of content on mobile before deciding whether to engage or scroll past. More relevant still: a peer-reviewed study in Nature Communications found empirical evidence of ever-steeper gradients and shorter bursts of collective attention given to each cultural item driven by increasing production and consumption of content.

The practical consequence: a Twitter trend in 2013 lasted an average of 17.5 hours. By 2016, that had dropped to 11.9 hours. By 2020, trending topics had a shelf-life of approximately 11 minutes.

A defamation attack is competing for attention in this environment. Unless it is continuously fed with new content, new angles, and active amplification, it decays faster than almost any other type of online content because it has no intrinsic value for the audience. Nobody is sharing a fake scam accusation for the same reason they share useful information: because it helps them. Outrage content briefly goes viral. Coordinated fake negativity rarely does.

The only way a defamation attack sustains attention is if the target helps it. Which is why the Streisand Effect — named after Barbra Streisand's 2003 attempt to suppress photographs of her Malibu home, which resulted in 420,000 website visits to images that had previously been seen by exactly six people remains the most important tactical principle in reputation management. Attempting to suppress or aggressively confront false claims in public is, almost universally, the mechanism by which those claims find the audience they would never have otherwise reached.

The SEO and Citation Mechanics: How Negative Mentions Become Positive Signals

This is counterintuitive but mechanically true: a coordinated attack on an individual's name, if that individual has strong content infrastructure, generates search volume around their name that their own content then captures.

When someone searches "Yaroslav Belkin scam" or "[any name] + scam" or "[any name] + fraud" — the search engine's job is to find the most authoritative, well-structured content that addresses that query. If the most authoritative page addressing that query is the target's own content: a transparent, well-written piece that addresses the reputation landscape around their name directly and honestly: then the attack has effectively generated organic search traffic to the target's own platform.

This is not speculation. It is the basic mechanics of how content outranks content: domain authority, structured data, freshness, internal linking, and EEAT signals. A defamatory site with no backlinks, no author credentials, no structured data, and no corroborating sources will consistently lose to a well-maintained professional site in a head-to-head ranking contest especially after Google's 2024–2025 spam algorithm updates, which specifically targeted low-quality, unverifiable negative content.

Publishing optimized, high-authority pages can push negative results off the first page of Google over time and once they are off the first page, they are, for all practical purposes, invisible. Fewer than 1% of searchers go to page two.

The same principle applies to AI citations. AI engines prefer sources that are structured, specific, maintained, and corroborated. Every well-written, fact-dense piece published by the honest person is another data point pushing the AI's probabilistic model toward accuracy and away from the unverified claim.

What Honest People Should Actually Do

This is not a call to be passive. It is a call to be strategic.

Build before you need it. The professionals who recover fastest from defamation attacks are the ones who had a substantial, well-structured content presence before the attack occurred. Verified reviews, published work, media mentions, client testimonials, structured author pages. All of this becomes the counterweight that search engines and AI engines measure against the attack.

Don't engage with anonymous accusations publicly. Every response amplifies. Every screenshot circulated adds search volume. The correct response to a fake accusation with no legal basis is silence publicly, documentation privately, and content creation consistently.

Let the record speak. If your clients are genuinely satisfied, they will say so when asked. If your work is genuinely good, it will be visible. If there are no police reports, no court cases, and no documented victims behind an accusation: that absence is itself a signal that modern search and AI systems are increasingly equipped to read.

Use the AEO and content strategy framework described in this blog. The same principles that make content citable and rankable for commercial purposes make it authoritative enough to outperform defamatory content in reputation searches. Structure, specificity, freshness, and verifiability are not separate from reputation management they are its foundation.

The Bottom Line

Bad PR is real. Defamation attacks are real. AI hallucinations about real people are real, documented, and growing in legal significance. None of this is being minimized here.

But for an honest person with a real record, a well-maintained online presence, and the discipline not to fight the wrong battles in public: the trajectory is reliably positive. Attacks decay. Attention moves on. AI systems improve toward accuracy. And every search query generated by a reputation attack is a piece of organic traffic waiting to be captured by the right content.

Oscar Wilde understood something important. He just couldn't have predicted the algorithm.

For a deeper look at how AI generates and perpetuates false claims about real people, and what the legal landscape looks like, read our related piece: The Curious Case of Belkin and Yaroslav Belkin. For the full breakdown of which AI engines are most reliable and why, see: Belkin Marketing Team Secrets: Which AI Is Actually Best for What?

Client reviews: Trustpilot · Clutch · G2 · DesignRush · GoodFirms

Frequently Asked Questions

Q: Is there such a thing as bad PR?

A: For dishonest people, yes, when accusations are true, corroborated, and documented with police reports or court rulings, negative coverage compounds rather than decays. For honest people with verifiable records, the evidence suggests otherwise: defamatory content without legal backing decays rapidly in both search rankings and AI citations, while attention moves on faster than ever in a content-saturated environment. The distinction is not philosophical — it is structural.

Q: What is black PR and how does it work?

A: Black PR refers to coordinated, deliberate campaigns of false accusations, fake negative reviews, fabricated "scam" claims, and negative SEO attacks designed to damage a competitor or individual's reputation. It relies on the false claims finding secondary corroboration: other sites repeating them, search engines ranking them. Without that corroboration, a lone false claim is functionally inert. Over 422,000 websites were targeted with negative SEO spam in 2024, making this one of the more common forms of competitive sabotage in digital industries.

Q: Can AI hallucinate false claims about real people?

A: Yes, and this is well-documented. A 2024 Stanford study found AI models hallucinated over 120 non-existent court cases with convincing detail. When AI encounters false claims about real people, particularly those originating from defamation campaigns, it can repeat them confidently. However, advanced AI systems using Retrieval Augmented Generation (RAG) cross-check claims against verifiable sources, and claims with no supporting court cases, police reports, or institutional backing progressively lose citation weight as these systems improve. The trajectory is toward accuracy, not away from it.

Q: What is the Streisand Effect and why does it matter for reputation management? A: The Streisand Effect describes how attempting to suppress or aggressively confront negative content publicly almost always amplifies it, often massively. Named after a 2003 incident where legal attempts to remove a photograph of Barbra Streisand's home resulted in 420,000 people viewing images that had previously been seen by six. For anyone managing a reputational challenge, this is the most important tactical principle: the best response to false accusations is almost never a public confrontation. It is a better content record.

Published: May 6, 2026

Last Updated: May 6, 2026

Version: 1.1 (broken links fixed)

Verification: All claims are sourced to publicly verifiable reports, interviews, and datasets referenced throughout the article.