Publicis CEO Niel Bornman calls AI slop “the newest battleground”

Publicis Groupe Connected Media CEO Niel Bornman is blunt about the moment: AI slop has become “the newest battleground” for brands competing online. Fake reviews and AI-powered search engines now provide instant responses that keep users from clicking through to official websites. Bornman says some brands have seen organic search traffic decline by between 5% and 35% as AI answer engines supply the answers before the visit.

That traffic squeeze feeds a second problem that’s harder to quantify but easier to feel. Over half (53%) of consumers distrust AI-generated search results and summaries, according to a 2025 survey by Gartner. And most people (70%) are uncomfortable with AI-generated media, a separate global survey by the management consultancy Baringo reported. So brands are stuck between two imperfect options: avoid AI slop headlines and skepticism, or use AI tools to “feed the machine” and remain visible in rankings. Bornman frames it as a dilemma that’s simultaneously operational and reputational: “Brands desperately want to avoid an AI-slop scandal,” but still need to appear as answers to consumer questions.

If you’re wondering how we got here, the source of the mess is also the business model. “AI slop” is the broad term for shoddy, algorithmically generated content flooding the internet, from rip-off books and fabricated reviews to content that is increasingly indistinguishable from reality. That last part is what turns the problem from annoying into dangerous. When fake and real look alike, people start acting like detectives. And when the web becomes a constant authenticity test, attention gets more expensive.

Bornman describes the cost dynamic in plain terms. Brands are being pushed to spend more on pay-per-click advertising while simultaneously feeling pressure to use AI themselves to produce content at the scale needed to stay visible. In other words, the same systems that amplify slop are also where brands must operate, meaning the cleanup effort is happening inside the same ecosystem that creates the issue.

This also explains why executives are moving in parallel across industries, even without agreement on what “good” looks like. LinkedIn, for example, announced a crackdown on “generic” content that “lacks authenticity and originality,” while rolling out generative AI features, including a “rewrite with AI” button embedded directly into its post composer. That kind of contradiction is the new normal: platforms want to restore trust but also monetize participation in AI-assisted production.

Publishing is the sector where the anxiety has become visible in a high-profile way. In March, Hachette withdrew the novel Shy Girl following allegations that sections of it had been generated using AI. The author denied directly using the technology, claiming instead that an editor inserted machine-generated passages into an early draft. Semantics matter, but the signal is larger: the industry’s ability to identify AI-generated material in manuscripts is under strain. Dan Conway, chief executive of the Publishers Association, calls it “the Wild West.” He argues that large language models are “hoovering up everybody’s content” and using it with reckless abandon, and he points to the mismatch between harmless inaccuracies and high-stakes domains like medical or educational material.

So what are the proposed fixes? One approach is economic: make it harder to profit from AI slop. Conway says if slop can’t be monetized, the incentive to produce it relentlessly decreases. The other approach, increasingly becoming its own market, is technological detection and verification. Companies are investing in AI-detection tools and verification systems designed to distinguish authentic work from machine-generated content, including tracing where online material originates. Pinterest has introduced labels for AI-generated images. Spotify has reportedly removed millions of bot-generated tracks using spam-detection systems. Bornman notes an “explosion” in AI-detection technologies, along with a practical boardroom question: what does this mean for my brand, and do we need to invest?

Regulation is also moving the timeline forward. From December 2026, the European Union’s AI Act will require many forms of AI-generated content to include digital watermarking, hidden digital markers indicating material was created using AI. More futuristic systems are also emerging, including blockchain-based provenance tools intended to maintain verifiable records of how digital content was created and modified. Some tech leaders compare this shift to cybersecurity in the early 2000s, where antivirus tools tried to filter malicious programs. Mel Morris, chief executive at Corpora.ai and described as the Candy Crush CEO in the source, makes the analogy explicit: detection tools are meant to stop unwanted content entering the system.

But detection has a downside that is business-critical: it is not an exact science. Most tools cannot definitively determine whether something is human- or machine-made; they estimate the probability that content might be AI-generated. That creates false positives, where human work is incorrectly flagged. Wipro’s global chief privacy officer, Ivana Bartoletti, highlights a “serious discrimination issue” in that context. She is not a native English speaker and says her writing tends to be structured, concise, and bullet-point heavy. When she runs her own work through AI detection systems, she says it is frequently flagged as machine-generated. Her concern is that neurodivergent people, non-native English speakers, or those who write in a more formulaic way could be unfairly penalized in hiring or corporate environments, not for using AI but for having a communication style that resembles what detectors expect.

Even the line between acceptable assistance and “slop” is fuzzy. Bartoletti asks: what happens if something is 5% AI-generated? Is that acceptable, or slop? What about 50%? It’s no longer black-and-white. Ironically, Bornman also worries about a second-order effect: systems meant to restore trust online may make the internet feel even more exhausting and suspicious. If detection and verification become part of everyday browsing, users will be asked to verify authenticity themselves, turning ordinary online activity into a constant exercise in doubt.

For executives, the strategic stake is straightforward: the brand problem is no longer just about creating content. It’s about surviving a trust economy where visibility, authenticity, and verification tooling are now entangled. In the same environment where content generation is cheaper and faster, the cost of being wrong, misread, or over-flagged can rise quickly. And the tools meant to restore confidence may also add friction that changes how people engage with brands in the first place.