Tyler Shultz helped bring down Theranos. Now he’s warning that the AI boom could produce the next one. TechCrunch AI reports that Shultz, in an interview with The San Francisco Standard, called the current AI frenzy “a great opportunity to go deceive people,” because it’s so hard for investors to tell real breakthroughs from clever packaging.
The timing matters. “You Can See Everything,” the Nathan Fielder and Lance Oppenheim documentary about Elizabeth Holmes’s last weeks before prison, is now one of the highest-grossing documentaries ever. It shows Holmes and her partner Billy Evans working on Haemanthus, a diagnostics company they call “Theranos 2.0.” It has reportedly raised about $20 million.
Why Shultz’s warning carries weight
Shultz isn’t a pundit throwing out hot takes. His first job out of college was at Theranos, and he was a key whistleblower who exposed its inaccurate blood tests. Those tests ended up in Walgreens stores and gave faulty results to real patients. Theranos hit a peak valuation of around $10 billion even though its technology never worked. The company ran only a fraction of tests on its own machines and sent the rest to third-party equipment.
He was right once, early, and at real personal cost. TechCrunch AI notes that Theranos targeted him with harassment while he was trying to raise the alarm. That record earns his warning a hearing.
The due diligence gap
Shultz’s main point is simple: AI is hard to vet.
“It’s really hard to tell the difference between what is something truly unique and new versus like, a ChatGPT wrapper,” he said. A “wrapper” is a product that mostly puts a new interface on someone else’s model, like OpenAI’s or Anthropic’s, without much proprietary technology underneath.
The Theranos playbook relied on that same opacity. Holmes refused to share scientific evidence with investors and said she was protecting trade secrets. Top venture firms walked away. Wealthy families like the DeVos and Walton families didn’t. Shultz describes it as “the biggest angel round of funding of all time.”
“I think the days of having a thirty-minute coffee with someone and sending them $100,000 should be over,” he said. “Elizabeth took that model and scaled it to $100 million.”
Now add an AI market where companies that have never made a profit are raising money at trillion-dollar valuations. The pressure to get in fast is huge, and careful vetting slows you down.
A fair counterpoint
AI isn’t a sealed black box the way Theranos’s machine was. You can usually test the product yourself. Run it, benchmark it, try to break it. A blood analyzer locked behind trade-secret claims is a lot harder to check than a chatbot you can poke at in five minutes.
The risk sits elsewhere, though. It’s in the claims you can’t easily test: proprietary model performance, revenue quality, defensibility, and how much of the “AI” is actually people doing the work behind the scenes. Regulators have already noticed. The SEC brought its first “AI washing” cases against investment advisers in 2024 for overstating their AI capabilities.
What the next 1 to 3 years probably look like
- Diligence gets technical. Expect more investors to require hands-on access, independent evaluations, and technical advisors before they sign anything.
- The wrapper shakeout speeds up. As foundation models take on more features themselves, thin products lose their edge, and their valuations may fall with them.
- More AI-washing enforcement. Inflated AI claims are an easy target for regulators, especially once the market cools.
- Health AI gets extra scrutiny. With Holmes back in diagnostics, anyone selling AI-powered medical tools should expect harder questions.
What to do now
- Investors: Ask what’s proprietary. If the answer is “trade secrets” and nothing you can test, take that as a warning sign.
- Businesses buying AI: Pilot it on your own data before committing. A great demo isn’t proof.
- Founders: Be clear about what you built versus what you license. Being honest early beats getting caught later.
Holmes’s second act
Shultz says the only reason Holmes is back in the conversation is “because she wants us to be talking about Elizabeth Holmes right now.” He doesn’t know where her next venture is headed. “I have this feeling of unease about what might come next,” he said.
It’s an uncomfortable mix: a sympathetic documentary, a new diagnostics startup, and a market that rewards bold claims faster than it checks them. The lessons from Theranos are still fresh. The question is whether AI investors use them before the next blowup. You can find the full story at TechCrunch AI.