{"slug": "wtf-is-ai-poisoning", "title": "WTF is AI poisoning?", "summary": "Marketers are increasingly concerned about AI poisoning, where competitors manipulate LLM responses by posting fake reviews or misinformation to damage rival brands. A study found only 8% of users verify AI-generated answers, and 71% distrust brands when AI responses conflict with brand information. Brands are investing in content marketing and SEO to protect their AI search visibility, but black-hat tactics remain a threat.", "body_md": "Prices rise tomorrow. Last chance for best rate.\n\n[VIEW PASSES](https://digiday.com/events/digiday-publishing-summit-sep-26/?source=display&utm-source=display)\n\nThis article is a WTF explainer, in which we break down media and marketing’s most confusing terms. [More from the series →](https://digiday.com/series/wtf/)\n\nMarketers have been trying to game LLM responses to suit their brands since users began turning to AI for search.\n\nIn recent months, they’ve become aware that AI search responses aren’t just a user touchpoint that needs attention, but a competitive arena in which they might influence the way their rivals are perceived — and have to guard against their rivals doing the same to them.\n\nDespite the well-publicized issues with AI hallucinations and bias, web users tend to perceive AI results as neutral. When a response supplies positive information about a brand, that works in marketers’ favor. When an AI Overview or ChatGPT response goes in the other direction, it can become a serious problem for a brand.\n\nIt’s compounded by the fact that, according to one study, [just 8% of people double-check facts in AI-generated search answers](https://www.inc.com/jessica-stillman/are-you-too-trusting-of-ai-answers-92-percent-of-people-dont-check-it-for-accuracy/91209990). And a survey of 1,000 consumers published by content marketing agency Skyword found that, when the information given in an AI response differs from that provided by a brand, web users distrust brands.\n\nJust 29% said they’d take an advertiser’s word for something in the event of a disagreement between the sources, the survey found. “Trust is no longer owned by a company, it’s earned across a broader information ecosystem,” said Skyword CEO Andrew Wheeler.\n\nMarketers across the industry are concerned about the potential of AI-dealt reputation damage. “The accuracy of content online about our clients is a core challenge with AI,” said one holding company media executive, who exchanged anonymity for candor.\n\nBrands have spent the last couple of years investing in their organic activity on platforms like Reddit and YouTube, increasing the output of their content marketing and PR operations, and redesigning websites to influence AI scrapers. Reddit is a favorite battleground for marketers vying to influence LLM results, [though YouTube](https://digiday.com/marketing/the-rundown-why-youtube-has-become-key-for-brand-geo-strategies/), Facebook, Instagram are also potential vectors.\n\nTheir rivals can use the same techniques, however, as part of a black-hat SEO approach designed to “poison” an LLM’s responses regarding a rival brand, potentially ruining its profile among AI search users.\n\n“[Competitors] could fund or coordinate content across forums, review sites, comparison pages, sponsored articles, influencer posts or other third-party sources in a way that makes a competitor look worse or their own brand look better,” explained Charlie Marchant, CEO at SEO agency ExposureNinja. “If those sources are later crawled, indexed or retrieved by AI search systems, they could potentially influence how brands are represented.”\n\nAI poisoning might involve posting bad-faith product reviews or outright misinformation about a competitor in the hope it’s ingested by an LLM and later regurgitated as honest opinion or fact.\n\n“[Competitors] can’t really de-rank you in ChatGPT, but they can definitely trash your reputation,” warned Jordan Parkes, CEO of specialist research company ZeroClick Labs.\n\nRemedies range from proactive measures to counter-attacking response, guided by AI search visibility tools like Profound or Peec.\n\nOn the defensive side, practitioners recommend marketers simply dial up the same work they’re doing to improve their AI search visibility. That stretches from ensuring social and website content uses consistent accurate language, to producing larger amounts of content like product specifications or user guides. Product and service guides account for up to 28% of AI search citations, [per a ZeroClick Labs estimate](https://zeroclicklabs.ai/ai-seo-study-content-types-ai-search/).\n\nIn some cases, clients are “fighting fire with fire”, according to Parkes; wading into subreddit discussions or social media threads in order to correct the factual record in the hope that a swift rejoinder will be ingested alongside any potential misinformation.\n\nWhile Parkes highlights the risks for brands, other practitioners are skeptical that brands can coordinate enough content across enough platforms to effectively influence or alter the ways an LLM presents a rival.\n\n“It’s far harder than many people assume, particularly for established brands. AI models don’t rely on a single Reddit thread or isolated source … They aggregate signals from websites, reviews, news coverage, social platforms and other trusted sources,” noted Charlie Terry, founder and CEO of performance marketing agency CEEK.\n\nIt’s also difficult to distinguish between a bad-faith PR operation funded by rivals and genuine errors (or honest, negative reviews) that happen to be published online.\n\nTerry suggested that, rather than attempting to salt the earth for competitors (or worry about SEO conspiracies), CMOs should focus on improving the AI discoverability of their brand content. “The brands that perform best in AI-generated recommendations are usually those with the strongest digital reputation,” he concluded, “not those trying to game the system.”\n\n### More in Marketing\n\n####\n[\nAgency bosses say the AI gap with clients is only getting wider ](https://digiday.com/marketing/agency-bosses-say-the-ai-gap-with-clients-is-only-getting-wider/)\n\nThe more versions of this perspective that got shared, the more it became clear there was a deeper frustration underneath it from others.\n\n####\n[\nElectronic Arts is betting that in-game ads can out-earn CTV ](https://digiday.com/marketing/electronic-arts-is-betting-that-in-game-ads-can-out-earn-ctv/)\n\nTo make in-game ads stick, EA has built its own stack rather than rent one. Now it wants to shape the standards before anyone else does.\n\n####\n[\nFuture of Marketing Briefing: Why Bose is building an entertainment company ](https://digiday.com/marketing/future-of-marketing-briefing-why-bose-is-building-an-entertainment-company/)\n\nBose has a new entertainment division. Its CMO hasn’t used a creative agency in five years. The two things are related.", "url": "https://wpnews.pro/news/wtf-is-ai-poisoning", "canonical_source": "https://digiday.com/marketing/wtf-is-ai-poisoning/?utm_campaign=digidaydis&utm_medium=rss&utm_source=general-rss", "published_at": "2026-07-06 04:01:00+00:00", "updated_at": "2026-07-07 01:28:51.066015+00:00", "lang": "en", "topics": ["large-language-models", "ai-safety", "ai-ethics", "ai-products", "ai-tools"], "entities": ["Skyword", "Andrew Wheeler", "ExposureNinja", "Charlie Marchant", "ZeroClick Labs", "Jordan Parkes", "Profound", "Peec"], "alternates": {"html": "https://wpnews.pro/news/wtf-is-ai-poisoning", "markdown": "https://wpnews.pro/news/wtf-is-ai-poisoning.md", "text": "https://wpnews.pro/news/wtf-is-ai-poisoning.txt", "jsonld": "https://wpnews.pro/news/wtf-is-ai-poisoning.jsonld"}}