An AI-generated formula is neither certified safe nor guaranteed stable without testing and human review; the formulation itself is only the starting hypothesis. Safety and stability in cosmetics depend on three things: (1) the formula's internal chemistry (ingredient compatibility, pH, preservative efficacy), (2) real-world testing (microbiological challenge tests, stability tests under heat/light/humidity, and if needed, dermal/ocular irritation studies), and (3) regulatory compliance (ensuring that each ingredient is listed on the positive list for your target market, meets purity and concentration caps, and is approved for the intended use). An AI tool can score well on item (1)—it can ensure that your ingredients are chemically compatible, that your water-based emulsion has the right ratio of emulsifier to oil, and that your preservative system is appropriate for the formula's water content and pH. But items (2) and (3) require real work. Stability testing, at minimum, involves storing your formula at room temperature, 40°C/75% RH (a common accelerated protocol), and sometimes under UV light for 8–12 weeks, then checking for changes in appearance, odor, viscosity, phase separation, or active ingredient potency via HPLC. A formula that looks stable on paper may separate, discolor, or lose efficacy during this testing. Microbiological challenge testing—in which a known pathogen (e.g., Staphylococcus aureus, Candida albicans, or Aspergillus brasiliensis) is added to your formula and its survival rate over time is measured—is the gold standard for confirming that your preservative system actually works; it is legally required in most markets if you claim antimicrobial benefit, and strongly recommended for any rinse-off or leave-on product with significant water content. The cost of these tests ranges from roughly $2,000–$8,000 depending on the protocol depth and the lab. If your formula includes a new active ingredient, skin sensitization or irritation testing may also be needed, especially if you are selling in the EU (where stability and irritation testing is more routinely required by brand standards). None of this testing can be skipped, and none of it can be guessed from the ingredient list alone. The way founders typically handle this is: (1) generate an AI draft, (2) have a real cosmetic chemist review it for feasibility and cost, (3) the chemist scales the formula to a small batch (500 mL to 2 L) and runs a quick in-house assessment (does it emulsify, does it feel right, does the pH hold?), (4) based on that feedback, you either refine the formula further or move to formal stability testing at an external lab. AJ Cosmo Labs makes step (2) practical: you post your AI-generated formula as part of a hire request, and vetted formulators can review it, flag problems, and quote you a price for testing and optimization. This is far cheaper and faster than hiring a chemist without context, and it ensures that by the time you pay for formal testing, the formula has already been sense-checked by a professional.
Read the full guide on Is There an AI Tool That Can Create Cosmetic Formulations? →