What happens when a newspaper with global reach publishes a glowing feature on an AI company running a scam?
That's the situation described in a Techdirt investigation into a recent New York Times piece about an AI-powered telehealth company. The Times presented the company as evidence of healthcare's AI-driven future. According to Techdirt's reporting, the company was operating a fraudulent scheme - and the Times apparently published without uncovering it.
Why AI Journalism Keeps Getting This Wrong
The AI hype cycle has created a specific vulnerability for journalists: companies claiming capabilities they don't have, pitching to reporters who are eager to cover the technology but don't have the background to stress-test the claims. A demo that looks impressive in a 30-minute briefing can hide a lot.
Telehealth is especially fertile ground for this. The category sits at the intersection of two subjects editors love to feature: healthcare transformation and AI. A company claiming to use AI to improve patient outcomes and reduce costs sounds like exactly the kind of story that belongs in a major newspaper feature.
"Using AI" has become nearly meaningless as a claim. Every company says it uses AI. Many are wrapping off-the-shelf APIs from OpenAI or Anthropic in a thin interface. A few are fabricating capabilities entirely. Without access to the underlying system and a technical reviewer who can evaluate it honestly, the difference is nearly impossible to spot from a press briefing.
The Verification Problem
Verification is structurally hard right now. Demos are staged. Customer testimonials can be produced on request. The metrics companies use to prove AI impact - "70% more efficient," "cut processing time in half" - are almost impossible to independently confirm. A reporter on deadline, working in a field they're still learning, writing about a company with polished PR, is at a genuine disadvantage.
That's not an excuse for skipping the basics. A telehealth startup claiming to have solved problems that well-funded incumbents haven't cracked should trigger more questions, not a feature profile. The failure mode here is treating AI claims differently from any other extraordinary claim - too much awe, too little skepticism.
The cost goes beyond one bad article. Credulous coverage gives real legitimacy to bad actors, makes it harder for honest companies to compete against misleading claims, and contributes to a media environment where AI criticism gets dismissed as technophobia. Journalists who want to get AI coverage right have a clear model: treat the underlying claims as claims, not premises.