The Theranos whistleblower sees the same danger in AI that he saw in the lab

Tyler Shultz didn't need to see Nathan Fielder's You Can See Everything, the documentary about Elizabeth Holmes, to confirm what he already knew: the Theranos story isn't over, it has simply changed venues. When his interview with The San Francisco Standard circulates through social networks, the question that appears most is not about Holmes — it is about what happens when the same pattern repeats itself in a sector that no one can audit.

Theranos, which reached a valuation of $10 billion without its technology actually working, taught a brutal lesson to Silicon Valley investors. Holmes and her partner Sunny Balwani used trade secrets as a shield to avoid scientific due diligence, raised hundreds of millions from the DeVos family and patriarch Sam Walton, and delivered false laboratory results to real patients through Walgreens pharmacies. Shultz, who was an intern when he realized the Edwards device did not work, reported the company and helped topple a fortress that thought itself untouchable.

What worries him today is not Holmes — she is serving an 11-year sentence and doing what she has always done: keeping herself at the center of attention. "The only reason we are talking about Elizabeth Holmes right now is because she wants us to talk about Elizabeth Holmes right now," Shultz told The SF Standard. Holmes is preparing a film, a book, a kind of second act that, according to him, "I don't know exactly where it's going."

Shultz's real warning points in another direction. "A huge danger with AI is that it is really hard to do due diligence," he said. "It's really hard to tell the difference between something truly unique and new versus a ChatGPT wrapper. If you want to deceive people, this is a great opportunity to deceive."

What is Haemanthus and why Shultz is concerned

Billy Evans, Holmes's partner, founded Haemanthus in 2022, an Austin-based startup that claims to use lasers and artificial intelligence to detect diseases in blood, saliva, and urine samples. According to marketing materials analyzed by The New York Times, the company's device employs a laser to analyze biological samples "at a molecular level," identify biomarkers such as glucose and hormones in seconds, and use AI to detect cancer and infections.

The technique itself — Raman spectroscopy — was discovered in 1928 and is widely known by the scientific community. There is nothing scientifically novel here. The problem, for Shultz, is the pattern of presentation: trade secrets as justification for not sharing data with investors, promises of equipment miniaturization to a bench-top microscope-sized device (the same misstep Theranos made), and a missing CTO who, according to Fortune, "disappeared before the founding article was published."

Haemanthus itself denies any link to Theranos on its X profile: the phrase "Elizabeth Holmes has zero involvement" — in some way ambiguous — suggests exactly the opposite of what it intends. The company claims to have patents and investors but has not yet published any peer-reviewed article proving the device's efficacy.

Why Shultz's skepticism matters now

The context has shifted since 2018. By then, the collapse of Theranos had been the tech equivalent of the Enron scandal — a bright reminder that "trust me, bro" is not enough in biotech. Since then, the artificial intelligence sector has raised more than $1 trillion in global funding, with companies that have never turned a profit operating at valuations that defy physics and financial logic. The difference between 2018 and 2026 is not that fraudsters are more creative; it is that the noise is so loud that no one can distinguish signal from fraud from genuine innovation.

Shultz, who now works in cybersecurity, is not proposing panic. He is proposing a brake: "I think the days of having a thirty-minute coffee with someone and sending them $100,000 should be over." In 2026, with investment rounds valuing startups in the billions based on demos of models that no one can reproduce, the analogy he makes is not paranoid — it is factual.

The question to watch is this: when the next Shultz emerges, will he be able to be heard before the ball drops, or has the cost of auditing AI promises become so high that there is nothing left but to believe — or to distrust everything?

Sources: TechCrunch, Fortune, The Hill

✓ Independent sources cross-checked and verified before publishing