Ofer Familier, Co-Founder and CEO of Dig.
getty
Most brand reputation playbooks assume that a crisis begins with a bad product or tone-deaf campaign, but it can also start with synthetic accounts spreading a rumor or amplifying authentic but negative opinions on a topic your brand has no problem with, on a platform your marketing team isn't monitoring closely. A handful of posts can snowball into mainstream coverage before anyone notices the smoke.
In the old, text-led world, a coordinated attack required brute force, where hundreds of synthetic accounts spread a rumor and interacted with each other. These were expensive, slow to produce and easy to detect. But the video world is different, with coordinated attacks now combining synthetic with authentic, running multi-modal across sources, mutating daily and rarely showing up as a single event.
Generative AI has driven the shift. Online deepfakes increased from roughly 500,000 in 2023 to about 8 million in 2025. With AI tools such as synthetic voice-overs, AI-assisted video formats and AI-generated personas lowering the barrier to creating believable synthetic content, a reputation attack no longer requires organizational resources or technical sophistication. A handful of well-placed videos reaching the right audience can set the feedback loop in motion that humans carry forward.
The first generation of fake engagement used bots to inflate numbers. Today's campaigns are more precise. They start inside a specific community that's likely to react strongly. A small network of fake accounts posts around the same theme, generating concentrated early engagement that looks organic. Recommendation systems learn locally before they scale; concentrated engagement in one pocket signals relevance, and the platform pushes the content to adjacent audiences.
From there, psychology takes over. Consumers see the same topic several times in a day, from different creators, in different formats and all pointing in the same direction. People mistake feed saturation for public reality, and thus a small, engineered push turns into apparent consensus.
Most organizations use reputation monitoring tools built for a text-first internet that track mentions, keywords and hashtags, but culture no longer lives in static text. Perception now forms through short-form video and recommendation feeds that determine what millions of people see next.
Attackers know not to tag the brand or use easily traceable keywords. The signals that narrative spreads, like tone of voice, visual context, community behavior and creator credibility, are largely invisible to keyword-based systems.
The only way to monitor content in today’s video-first world is by moving from tracking words to tracking narrative sentiment. Brands can't anticipate every term an attacker will use, but they still need the tools to detect and stop attacks before they do meaningful damage. AI-equipped attackers in 2026 can’t be fought with tools from 1990.
An example of this played out when videos circulated claiming major luxury brands misled consumers about product origins and pricing. The narrative started with a small number of posts framed as insider scoops. Within days, creators built their own content on top of it, and our company's internal analysis observed the narrative spreading across 6,589 creators who published 7,888 videos that collectively generated approximately 1.15 billion views. Our investigation also found that only 40% of the accounts involved had any prior posting history in Mandarin or from China-based locations, despite the content being framed as coming from Chinese factory insiders.
The mismatch alone undermined the authenticity of the "insider" framing. The content itself told a similar story: our analysis found that most of the posts violated at least one TikTok policy, with nearly half promoting counterfeit or unauthorized sales, while others used deepfakes or inauthentic user information and showed signs of fake engagement or bot activity. The vast majority of posts pushed the same four coordinated narratives, further confirming this was not organic consumer outrage but an orchestrated attack.
Video makes content fast to consume, and that speed is what makes narrative attacks hard to contain. Creators don't need to be part of the original coordination to fuel amplification. They see accounts gaining outsized reach on a topic, conclude audiences want more and produce reaction videos: the chain continues.
Many executives still treat fake and authentic as binary states where a narrative is either manufactured or real. But a narrative that begins with synthetic behavior produces authentic consequences once enough real people adopt it, react to it, parody it and spread it through their networks.
With GenAI, people can only distinguish human content from AI-generated content 51% of the time. Once a real audience accepts a narrative, its origin becomes irrelevant. Brands usually take notice too late, after a rumor has spiraled from TikTok into customer complaints, journalist inquiries and investor demands. The signals are visible, just in places where organizations are not equipped to read.
Reputation crises are manageable in their early stages, but exponentially more costly as they grow. The real competitive advantage comes with seeing the problem earlier.
It helps to understand that these attackers built their tradecraft inside geopolitical conflicts and are now also targeting consumer brands that can offer a potentially massive financial windfall. The toolkit already exists; they just need a name to attack.
That makes this a board-level operating problem, not a marketing fire drill. Brands need to manage it end to end, with clear ownership of narrative risk, response protocols modeled on cyber threats, the speed to act once a story is forming, the methods for countering without amplifying and a budget that funds it all.
It all starts with coverage. You can't manage what you can't see, and you can't see what your text-based tools weren’t built to read. Brands need to know what story is forming, its origin and whether the momentum is organic or coordinated, because counting mentions and scanning comments isn't enough. The brands that get this right will treat narrative intelligence the way they already treat cybersecurity.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。