Outer Biosciences Is Training an AI Model on Skin That's Still Alive

No "biology internet" to scrape — so the startup built its own closed-loop system for AI-driven dermatology

Published: 2026-08-22 Category: Quick Take Sources: TechCrunch — Michael Polansky is training an AI model on skin that's still alive

What Happened

Michael Polansky — co-founder and CEO of Outer Biosciences, and the creative, business, and romantic partner of Lady Gaga — broke years of silence to reveal what his startup has quietly been building. Outer Biosciences keeps living human skin alive outside the body for over a month, far beyond the industry norm of days, then feeds the data from that living tissue into a closed-loop AI model.

The company sources human skin that would otherwise be discarded after surgery (mostly plastic surgery) through vetted non-profit and commercial biobanks operating under IRB oversight and documented donor consent. Samples arrive de-identified, and a proprietary support system feeds the tissue nutrients and removes metabolic waste. That extra viable time lets researchers track slow biological processes — collagen remodeling, pigmentation change, sunburn recovery — that take weeks to unfold.

Why It Matters

The core bet is that biology needs better data than animal models, simple cell cultures, or lab-grown organoids can provide. Those proxies are poor stand-ins for how a real human organ behaves. Outer's closed self-enforcing system works like this: an AI predicts which untested chemicals are likely to benefit a specific skin function; those chemicals run through the living-tissue system; and the results, whether right or wrong, feed back into the model to improve the next round of predictions.

The results are striking. Early on, a "brute force" mining of scientific literature produced a couple of leads over 18 months. With AI layered in, the company now generates a new candidate roughly every six weeks, with six leads active and several dozen additional "hits" logged. It's discovering cosmetic ingredients (not drugs), so there's no FDA approval to seek — instead the route runs through standardizing ingredient names and OECD safety testing.

The Deeper Signal

Polansky frames the company's real moat with a telling phrase: unlike AI companies training on scraping the internet, "there is no 'biology internet' to scrape." The data Outer Biosciences generates doesn't exist anywhere else. That makes the position more defensible — and slower to build — than a traditional software-based AI startup. It's also cheaper to run; all AI work stays on-premise because, as Polansky puts it, "we don't want the data in the cloud."

The ambition isn't just cosmetics. The hope is a predictive model accurate enough to spot promising directions in skin biology without running every experiment physically first — opening a rate of discovery in dermatology that doesn't currently exist. Polansky, who spent over a decade alongside Sean Parker in cancer immunotherapy, is explicit that the whole thing grew out of frustration that biology and chemistry have never moved as fast as software, because there's no ethical way to run experiments directly on people.

There's also a warning in here about the limits of the AI boom's data moats. In software, data is commoditized and abundant; in biology, the bottleneck is generating proprietary ground-truth data in the first place. Companies that can generate their own clean, real-world biological data — even slowly, even expensively — may end up with far more durable positions than the ones that just trained on the public internet. That's a model for what defensible AI looks like outside the text-and-code world.

Based on reporting by TechCrunch.