Midjourney, the AI startup famous for turning text prompts into glossy images, just announced it’s getting into medical imaging. The plan, according to The Verge AI, is a futuristic ultrasound scanner that lowers you into a vat of water and produces internal scans the company hopes will be “something as powerful as MRI” yet “as casual as a trip to the spa.” CEO David Holz has even suggested the system could one day beat MRI. Medical imaging experts who spoke to The Verge had one consistent response: show us the evidence.
This is a significant move, and not just because it’s strange. A company that makes synthetic images is walking into the high-stakes, tightly regulated world of medicine. What stands out here is the gap between the pitch and the proof.
What Midjourney is actually proposing
The machine at the center of “Midjourney Medical” is an ultrasound scanner. You stand on a platform, get submerged in water, and a ring of underwater sensors fires sound waves into your body and reads the echoes that bounce back. Midjourney compares it to dolphin echolocation and says a full scan should take no more than 60 seconds. Typical ultrasounds run 30 minutes or more. MRI scans take longer still and trap you in a narrow, noisy tube.
The AI angle is quieter than you’d expect from an AI company. Midjourney’s head of medical, Tom Calloway, told The Verge the scanner uses AI and specialized chips to process “unthinkably huge amounts of data,” plus lossless compression to speed things up. Notably, AI is barely mentioned in the company’s own blog post laying out the plan.
Here’s the clever regulatory sidestep: Midjourney isn’t calling this a diagnostic medical device. That would trigger FDA clearance and clinical trials. Instead, it’s framing the scanner as a wellness product that gives you “more information about your bodies,” with machines placed in spas where scans become “a side-effect” of a visit. The concept images show golden rooms and luxury pools, not a clinic.
Why the experts are skeptical
The specialists The Verge interviewed weren’t dismissive. Several called the concept genuinely exciting and even plausible. But they raised the same three issues:
- It’s not as novel as claimed. Ultrasound has been a diagnostic tool for decades, and researchers have already built whole-body prototype systems using related methods.
- The MRI comparisons aren’t supported. Venkatesh Murthy of the University of Michigan said Midjourney’s resolution claims are “clearly theoretical” and that calling it an MRI equivalent is “completely unsupported.” The images shown so far, he said, “are decidedly low-res,” despite the blog post promising detail down to a fraction of a millimeter.
- Net benefit to patients is unproven. Mark Anastasio of Washington University in St. Louis called it an “exciting development” but said there “is no current evidence” that detailed scans like this deliver clinical value yet.
Midjourney also quotes a striking statistic: the world could avoid “30% of all deaths and 50% of all healthcare costs” with enough early imaging. That’s the kind of number that sounds great in a launch post and means little without validation.
Why this matters now
Notice who cheered. Most of the online enthusiasm, The Verge reports, came from outside medicine and radiology. That split is the real story. Silicon Valley loves a disruptive moonshot, and “constant cheap body monitoring” is an easy sell to people who don’t work in imaging. Clinicians, who do, asked for data.
The wellness framing is the part worth watching. By skipping the FDA’s diagnostic path, Midjourney can ship faster and market harder, while patients may still read those spa scans as medical truth. That’s a familiar pattern in consumer health tech, and it tends to end in regulatory attention.
A few practical takeaways:
- For AI builders: moving into regulated fields means evidence is the product, not the demo. Polished renders won’t carry a medical claim.
- For businesses eyeing health AI: the wellness-not-diagnostic loophole works until it doesn’t. Plan for the clinical bar, not around it.
- For everyone else: treat any “better than MRI” claim as a hypothesis until peer-reviewed results land.
Midjourney says it plans to expand into real medical applications later. Whether that happens depends entirely on the answer to the question every expert asked: where’s the proof? You can read the full reporting at the original source.