// AI NEWS · MEDIA INTEGRITY

AI-Generated "Experts" Are Being Quoted by CBS News, the NY Post and Business Insider — Then Exposed as Fake

A wave of fabricated "subject-matter experts" — AI-written pitches behind AI-generated headshots — has been quoted as real by major newsrooms. Journalist-sourcing platform Qwoted is now banning dozens of fake accounts a day.

2026-09-23 · by Moe Ameen

What happened

Reporting through 2026 has documented a growing pattern: people who don't exist, invented as "experts" and pitched to journalists with AI-written commentary behind AI-generated headshots, getting quoted as authorities in major publications. The clearest case is "James Allen" of a site called Billpin, presented as a finance expert and quoted by outlets including CBS News, the New York Post, and Business Insider. Journalist-sourcing platform Qwoted flagged his account for repeated AI-generated submissions and a failed verification, and America's Financial Educators Council questioned his claimed qualifications. Afterward his LinkedIn profile vanished, Billpin's website went blank, and he never responded to requests to confirm he was real.

Allen is not isolated. A Press Gazette investigation named other suspected fabrications — "Grace Carter" of UKWritings, published across topics as far apart as ad tech and pet sitting, and "Dan Brown" of a cleaning company whose profile photo an image detector rated 9-out-of-10 likely AI-generated, yet who appeared in dozens of mainstream outlets. An earlier Press Gazette review of 250 expert-quoting articles across UK tabloids found roughly two dozen sources that don't exist, can't be verified, or were never actually named.

The motive is search visibility, not journalism. A quote in a respected publication typically carries a backlink and a credibility signal, so an obscure site has an incentive to manufacture a plausible-looking expert who can earn that coverage. Qwoted says it now handles upwards of 15 violations a day and blocks around 50 prospective sign-ups daily, using a human vetting team plus AI detection (it has cited the detector Pangram) to screen high-frequency, machine-written pitches. Its CEO Dan Simon put it bluntly: "AI has just poured petrol on the disinformation fire."

Two caveats travel with the story. Detection is probabilistic — an AI-likelihood score on a headshot is a strong signal, not proof — and some initially listed names were later contested or clarified. The verified core is narrower than the alarm: real newsrooms have quoted sources that turned out to be fabricated, and the platforms in the middle are now fighting it as an ongoing, whack-a-mole problem.

Why it matters for creators

  • Trust is the currency being counterfeited. As fabricated experts get caught, editors and audiences grow more skeptical of every unfamiliar name — which raises the bar for legitimate creators and experts to prove they're real, qualified, and consistent.
  • The whole scheme exists because earned media is valuable. That's a reminder of the actual prize: genuine authority and citations. The fakes are a shortcut to something you can build honestly by publishing substantive work at cadence.
  • AI-generated headshots and AI-written pitches are now cheap enough to mass-produce, so the differentiator shifts to verifiable identity and a real track record — a consistent face, voice, and body of published content tied to a real person.
  • Disclosure norms are tightening in parallel (platform likeness rules, EU labeling), and the line that matters is deception: using AI to scale your own real voice and likeness is fine; inventing a person to fool a journalist is the thing getting banned.
  • AI detectors and sourcing platforms are becoming gatekeepers. Generic, machine-written filler gets flagged; specific, on-brand content that reads like a real practitioner is what survives the screen.

How to act on this with Kompozy

Start with the boundary, because it's the whole point: [Kompozy](/) is not a tool for manufacturing a fake expert. It does not invent people, fabricate credentials, or auto-pitch journalists behind a synthetic headshot — that's exactly the behavior Qwoted is banning, and building on it is a reputational time bomb. Kompozy is the opposite play. It's for a real person or brand who already has expertise and needs to prove it at the volume that earns genuine authority, using their own likeness and their own voice, disclosed as AI-assisted where the rules require it.

The reason the fakes exist is that a credible, consistent expert presence is worth manufacturing. So build the real one. Kompozy's AI Influencer persona is your own face (Gemini face-lock keeps it consistent) and your own voice profile, governed by a Persona Brief and a banned-word filter so the output reads like you, not like the generic AI filler detectors catch. From one set of source material it generates Persona Shorts and avatar video, Carousel Posts, Blog Articles, Email Newsletters, and native Text Posts, then [Autopilot](/glossary/autopilot) and a per-post review pipeline schedule and publish the batch across the eight social platforms plus blog and email. That cadence — real substance, one identity, published every week — is how you become the source worth quoting instead of the one that gets unmasked. For the ethics of owned synthetic personas versus deceptive ones, see [the AI influencer manipulation trend](/guides/ai-influencer-manipulation-trend) and [AI virtual actors and influencers](/guides/ai-virtual-actors-and-influencers); for where disclosure lines are being drawn, see [YouTube's AI disclosure and likeness rules](/news/youtube-ai-disclosure-likeness-rules-2026).

Quick takeaways

  • "James Allen" of Billpin was quoted as a finance expert by CBS News, the NY Post, and Business Insider before Qwoted flagged AI-generated submissions and a failed verification; his LinkedIn and site then disappeared.
  • Press Gazette documented other suspected fabrications (Grace Carter, Dan Brown) and found ~24 unverifiable sources in a sample of 250 expert-quoting UK tabloid articles.
  • The motive is SEO: a quote in a real publication carries a backlink and credibility, creating an incentive to invent quotable "experts."
  • Qwoted says it handles 15+ violations and blocks ~50 sign-ups a day, using human vetting plus AI detection (e.g. Pangram). Detection is probabilistic, not proof.

Frequently asked questions

Who is James Allen of Billpin?

"James Allen" was presented as a finance expert affiliated with a site called Billpin and was quoted by major outlets including CBS News, the New York Post, and Business Insider. The journalist-sourcing platform Qwoted flagged his account for repeated AI-generated submissions and a failed verification, and a financial-education body questioned his qualifications. His LinkedIn profile then vanished, Billpin's website went blank, and he never confirmed he was a real person — the pattern of a fabricated identity built to earn media mentions.

Why would someone create a fake AI expert?

Search visibility. A quote in a respected publication usually comes with a backlink and a credibility signal, which can lift an otherwise obscure website in search results. That value creates an incentive to manufacture a plausible-looking expert who can earn coverage. It's a shortcut to the earned authority that a real, consistently publishing expert builds honestly.

How are platforms catching AI-generated fake experts?

Journalist-sourcing platforms like Qwoted combine human vetting teams with AI detection tools (it has cited the detector Pangram) to flag high-frequency, machine-written pitches and unverifiable identities. Qwoted says it handles more than 15 violations a day and blocks around 50 prospective sign-ups daily. Detection is probabilistic — an AI-likelihood score on a headshot is a strong signal, not definitive proof — so the fight is an ongoing whack-a-mole rather than a solved problem.

Is using AI to build an expert presence the same as faking one?

No — the line is deception. Inventing a person who doesn't exist, with fabricated credentials and an AI-generated headshot, to fool a journalist is what's getting banned. Using AI to scale your own real voice and likeness — publishing more of your genuine expertise, disclosed where the rules require — is a legitimate workflow. Kompozy is built for the second: a real person's own face and voice, governed by a Persona Brief, publishing consistently, not a fabricated identity.

Related news

← All AI news · Get started →