Theranos whistleblower Tyler Shultz warns that the AI boom is ‘a great opportunity to go deceive people’

We have entered the season of the ghost feature. It is that time in a tech cycle where the distance between what a pitch deck promises and what the code actually does becomes a chasm. I have seen this movie before, and so has Tyler Shultz. If that name sounds familiar, it is because he was the one who pulled the thread that unraveled Theranos.

Shultz is back in the conversation lately, and his warning is simple: the current artificial intelligence boom is a playground for deception. While the technology is certainly more tangible than a single drop of blood doing a thousand tests, the market dynamics are eerily similar. We are seeing valuations hit the stratosphere before a single cent of profit is realized, and for builders, that is a dangerous place to live.

The Vaporware Template

In the early days of Theranos, the deception was not just about lying to investors; it was about the culture of 'faking it' until you make it. The problem is that in deep tech—whether that is biotech or large language models—faking it is significantly easier than making it. AI is the perfect cover because the average investor, and even many sophisticated VCs, do not actually understand how the sausage is made.

When you tell someone that your model has emergent properties or that it is approaching artificial general intelligence, you are using words that sound scientific but lack a concrete, verifiable definition. This is exactly where Shultz sees the risk. If you can hide your lack of progress behind a curtain of proprietary algorithms and black-box complexity, you can raise hundreds of millions of dollars without ever having to prove the math works.

The Founder's Dilemma

For those of us actually building, this environment creates a toxic incentive structure. If your competitor is claiming they have solved reasoning while they are really just using a hidden layer of human labelers or a massive script, they get the funding. You, trying to be honest about the limitations of your RAG pipeline or your inference costs, look like a laggard.

Shultz points out that the pressure to keep up with the hype cycle often forces founders to make small concessions. Maybe you exaggerate the autonomy of your agent. Maybe you cherry-pick the one out of a hundred responses where the AI did not hallucinate. But as Shultz learned at Theranos, those small concessions eventually become the foundation of the company. Once that happens, you are no longer a tech company; you are a narrative company.

Valuations vs. Reality

The numbers we are seeing right now are frankly absurd. We have startups raising at billion-dollar valuations with little more than a wrapper around a third-party API. When the cost of doing business is controlled by three or four major compute providers, and your revenue is non-existent, the math does not add up. Shultz is reminding us that when the music stops, the people holding the bag are usually the ones who believed the hype most fervently.

In the crypto world, we saw this with the 'everything is a blockchain' era. Now, everything is an 'agentic AI.' The terminology changes, but the behavior of the bad actors remains the same. They use complexity as a shield against scrutiny. If you ask too many questions about the underlying tech, you are told you 'just don't get the vision.'

Why This Matters for Builders

If you are a founder trying to build something that lasts, the Shultz warning should be your north star. Deception is a short-term strategy with a terminal debt. The moment you start deceiving your investors or your customers about what your AI can actually do, you lose the ability to actually fix the problems. You become too busy maintaining the lie to iterate on the product.

• Focus on utility over narrative: If your product requires a 20-minute explanation of the 'future of humanity' to justify its existence, it probably doesn't work.
• Transparency is a competitive advantage: In a market full of black boxes, being the founder who says 'here is where our model fails' builds long-term trust that capital cannot buy.
• Avoid the 'Theranos Trap': Never let the marketing team get more than two steps ahead of the engineering team.

"The AI boom is a great opportunity to go deceive people because there is so much money and so little understanding of the actual mechanics."

That quote, paraphrased from Shultz, should be pinned to the wall of every YC cohort. We are in a bubble of expectations. Some of these companies will genuinely change the world, but many are just Theranos with a better UI. The builders who survive will be the ones who didn't mistake a high valuation for a functional product.

The Takeaway

The lesson from Tyler Shultz is not that AI is a scam, but that AI is the current preferred vehicle for scammers. As a founder, your job is to stay grounded in the technical reality of what you are shipping. Don't let the trillion-dollar hype train force you into making promises your code can't keep. The truth always comes out eventually—usually right after the last round of funding dries up.

Read the original at TechCrunch AI →