Authoritas Asked 9 AI Models Who to Trust. The Results Should Worry Business Leaders.
Last month, Authoritas published research that should terrify every business leader who hasn’t thought about how AI represents their brand.
They tested whether fake experts - manufactured personas with planted press coverage - could fool AI systems into recommending them as authorities. The short answer: no. But the longer answer reveals something far more important about how AI decides who to trust.
The Experiment
Authoritas queried nine AI models (ChatGPT, Claude, Gemini, Perplexity, and others) with two types of questions:
- Topic-based: “Who are the leading experts on [subject]?”
- Name-based: “Is [specific person] a recognized authority?”
For the fake experts - personas created through planted quotes in tabloid press - zero appeared in topic-based recommendations. The AI systems simply didn’t surface them when asked who to trust.
But here’s where it gets interesting. When asked directly about specific names, the AI models failed 29.5% of the time, sometimes confidently inventing career histories for people who don’t exist.
What This Means for Business
AI doesn’t rank pages anymore. It recommends entities.
When someone asks ChatGPT “What’s the best solution for [problem]?” or Google’s AI Mode “Who should I hire for [service]?” - the AI isn’t searching. It’s recommending. And it’s recommending based on what it believes to be true about a brand.
Research shows that real authorities dominate AI recommendations because they’ve built what Authoritas calls “deep, corroborated brand entities” - consistent profiles, third-party validation, years of published work creating an interconnected web of trust signals.
As Laurence O’Toole, CEO of Authoritas explains, the future of SEO goes beyond tracking traditional rankings - it focuses on how brands are understood and recommended by AI, with real authority and a verified digital presence determining visibility.
One expert in their study achieved a perfect 10/10 consistency score across all test queries. The AI systems cited him more frequently, more prominently, and across a wider breadth of topics than any competitor. Why? Because he’d spent a decade building exactly the kind of entity architecture that AI systems trust.
The Algorithmic Trinity
This isn’t just about Google anymore. Brands now exist across what industry experts call the “Algorithmic Trinity” - Knowledge Graphs, Large Language Models, and Search Engines working together.
Andrea Volpini, CEO of WordLift, describes it this way in his Software Advice review: “Entity optimization for AI is exactly where the industry needs to go.”
Koray Tuğberk Gübür, founder of Holistic SEO Digital, goes further in his in-depth analysis: “You need a complete picture of a brand’s position across the entire Algorithmic Trinity (Knowledge Graphs, LLMs and Search), offering a clear roadmap for achieving a positive, accurate, and convincing digital narrative.”
The brands winning in AI aren’t optimizing for one platform. They’re building entity architecture that works everywhere AI looks.
The New Due Diligence
Here’s what’s changed: Before the next meeting gets scheduled, someone is asking AI about the brand.
Not Googling. Asking AI.
“Tell me about [company].”
“What’s [name]’s background?”
“Should I consider [brand] for this project?”
And AI will answer. The question is: what will it say?
If a digital presence is fragmented, contradictory, or thin - AI hedges. It says things like “claims to be” instead of “is.” It mentions competitors. It stays silent when it should advocate.
If a digital presence is consistent, corroborated, and comprehensive - AI states facts with confidence. It recommends. It becomes a 24/7 sales force.
Three Questions Every Leader Should Ask
- What does AI say about my brand when I’m not in the room? Ask ChatGPT, Perplexity, and Google’s AI Mode. Not vanity queries - the queries prospects actually use.
- Is my brand being recommended or ignored? Search for “best [category]” and “who should I hire for [service].” Is the brand in the conversation? Are competitors?
- Is AI confident or hedging? Look for phrases like “claims to be,” “according to their website,” or “one of the leading.” These signal weak entity architecture.
The Uncomfortable Truth
The Authoritas research contained one line that should keep marketers up at night:
“You can trick a model into recognising a name, but you can’t trick it into recommending that name ahead of genuine authorities.”
There are no shortcuts. AI systems have become remarkably good at distinguishing between manufactured credibility and earned authority. The brands that win in AI recommendations are the ones that have done the work - consistently, over time, across multiple trusted sources.
The good news? Most competitors haven’t started. The AI recommendation layer is still being written. The question isn’t whether to optimize for it - it’s whether a brand will be recommended, or recommended against.
The research cited in this article is from Authoritas: “Can You Fake Expertise in AI Search? We Tested 9 Models to Find Out” (December 2025).
Further Reading:
