How Outer Biosciences Uses Living Human Skin to Train AI

Outer Biosciences is a biotechnology startup founded in 2022 and led by Michael Polansky, who is also publicly known as Lady Gaga’s partner. Its central idea is to use living human skin as a longer-duration testing environment for cosmetic and skin-health research, then turn the resulting biological measurements into training data for AI. 358

The approach is ambitious, but the evidence currently available is mostly company-reported. Public materials do not yet establish independent replication of the tissue-survival claim, published model benchmarks, clinical outcomes, revenue, or finished commercial ingredients. 1120

Why Michael Polansky is involved

Available reporting connects Polansky’s background to applied mathematics, finance, and venture investing, including work associated with Bridgewater Associates, Sean Parker, and Founders Fund. That combination helps explain the startup’s emphasis on capital-intensive biology, proprietary data, and a potential licensing business. 56

Public sources do not provide enough detail to independently verify every biographical element sometimes associated with Polansky, including specific claims about cancer immunotherapy and philanthropy. Lady Gaga is closely associated with the venture and has helped publicize it, but reporting is inconsistent about whether she is formally a co-founder. The more specific account of the founding team names Polansky, chief scientist Kyung-Jin Jang, CTO Chris Hinojosa, and chief business officer Stanley King. 35

The tissue platform: discarded skin kept alive outside the body

Outer says it obtains consented, de-identified, full-thickness human skin that would otherwise be discarded after surgery. Its system supplies nutrients to the tissue and removes metabolic waste through a perfusion and bioreactor setup. The company says this can keep the skin viable ex vivo for more than 30 days, compared with roughly a week for conventional ex vivo skin in its own comparison. 21925

That extended window is important because some skin processes are not adequately captured by a short assay. Outer says it uses the tissue to study responses such as UVB damage, inflammation, collagen remodeling, pigmentation, and barrier repair. The claimed benefit is the ability to observe changes over time in native, full-thickness human tissue rather than relying only on isolated cells or engineered substitutes. 813

The “30+ days” figure should be treated as a company claim, not as a proven industry standard. One analysis noted that the public disclosure does not define exactly how viability is measured, how many samples reach the stated duration, or whether an outside laboratory has reproduced the result. 11

How the closed-loop AI process works

Outer’s proposed workflow links computational prediction to repeated biological testing:

1. The company exposes living skin tissue to a chemical compound.
2. It measures molecular, cellular, and functional responses.
3. An in-house model learns relationships between compounds and specific skin functions.
4. The model ranks untested molecules that may produce a desired response.
5. Those predictions are tested on additional living tissue.
6. The results—whether positive or negative—are fed back into the model.

This is a closed-loop discovery system: experiments generate data, the model uses that data to select new experiments, and the new results improve the next round of predictions. 816

The claimed moat is therefore not simply an AI architecture. It is the proprietary, longitudinal dataset created by running experiments on human tissue in the company’s own laboratory. More observations over a longer period could, in principle, give the model information that short assays miss. But that advantage depends on the data being large, consistent, diverse, and predictive outside the original experiments.

Reported discovery speed

Outer says it moved from finding only a few leads during its first roughly 18 months to producing about one new candidate every six weeks. Reports also describe six active leads and several dozen hits. 21214

Those figures are not independent performance benchmarks. They do not, on their own, show how many candidates were tested, how often predictions were correct, whether the compounds work in human use, or how the company compares with established discovery workflows. The more meaningful test will be whether candidates survive safety, reproducibility, formulation, and real-world efficacy evaluation.

Consent, privacy, and oversight

Using tissue that would otherwise be discarded does not eliminate ethical or privacy obligations. Outer’s stated model relies on donor consent and de-identification before tissue reaches the company. Reports also describe sourcing through tissue banks, brokers, and other intermediaries. 1319

The practical questions are more specific: What exactly do donors consent to? When is institutional review board review required? How are tissue samples linked to metadata? Who can access the resulting biological data? And what safeguards prevent re-identification?

Public reporting does not provide enough detail to assess Outer’s consent forms, IRB protocols, data-governance procedures, or re-identification controls. De-identification reduces direct identifiers, but tissue-derived data can still require careful handling when combined with other information.

Cosmetic research is not the same as drug development

Outer’s initial focus appears to be cosmetic ingredients and skin-health research rather than therapeutic disease claims. That distinction affects what evidence and regulatory pathway a product may need.

A promising result in an ex vivo human-skin assay does not automatically replace required safety testing or establish that an ingredient is effective in consumers. Existing alternative-testing frameworks include reconstructed-human-epidermis methods and integrated in vitro, in chemico, and in silico approaches covered by OECD guidance. 1718

The company’s platform could become one part of a broader evidence package, but its assay alone should not be interpreted as regulatory approval or a substitute for every required test.

Business model and funding

Outer appears to be building a business around licensing or commercializing validated ingredients, assays, and discovery programs for beauty, skincare, pharmaceutical, and ingredient-supply partners. Its own partnerships page describes work with biotech, pharma, skincare and cosmetics companies, and ingredient suppliers. 2627

Public reports put disclosed funding at approximately $23 million. They do not establish revenue, signed partner contracts, or a commercially launched ingredient. 1420

That model would let Outer sell access to biological insights and discovery capabilities without becoming primarily a consumer skincare brand. The company’s commercial value will ultimately depend on whether its experimental system produces compounds that partners can formulate, validate, and bring to market.

How Outer differs from organ-on-a-chip companies

Outer is part of a wider effort to develop more human-relevant alternatives to conventional laboratory and animal-testing approaches. Organ-on-a-chip, organoid, and other microphysiological systems use engineered tissues or cell-based constructs to model human biology and can support safety evaluation and product development. 1824

Outer’s stated differentiation is different: it emphasizes prolonged experiments on donated, native, full-thickness human skin, combined with an on-premises AI model trained on the resulting data. 416

That is a distinction in experimental substrate and data strategy, not proof that Outer has made physical experiments unnecessary. Its long-term ambition is to build a model capable of prioritizing—or eventually simulating—more experiments before tissue is used. For that to be credible, the model would need to generalize across donors, skin tones, anatomical sites, chemical classes, and real-world formulations.

What to watch next

The most important question is not whether Outer can keep a piece of skin alive for a month. It is whether the platform can generate predictions that remain accurate when tested on new donors and compounds, and whether those predictions lead to safe, reproducible ingredients with measurable human benefit.

The company’s strongest potential advantage is the feedback loop between living tissue and machine learning. Its largest unresolved risks are also concentrated there: limited public validation, unknown dataset scale and diversity, unclear oversight details, and the gap between an interesting laboratory signal and a commercially useful product. For now, Outer Biosciences is best understood as a promising but early attempt to make AI-driven skin-ingredient discovery more biologically grounded.