If an AI discovers a drug, who gets the money? The answer hinges on intellectual property law, contractual agreements, and how human involvement is documented. While AI can generate promising molecular candidates quickly, U.S. patent law currently does not recognize AI as an inventor. A 2022 court ruling, Thaler v. Vidal, affirmed that only humans can be named inventors on patents. Although AI’s capabilities have advanced significantly since then, legal precedent is slow to change, and the U.S. Supreme Court would need to overturn the decision for that to shift. The U.S. Patent Office has issued new guidance treating AI as a tool—like a computer or lab instrument—rather than a creator. This means that for a patent to be granted, there must still be a human inventor who contributes to the conception of the invention. In drug discovery, this often means the chemist who figures out how to synthesize a molecule suggested by AI may still be considered the inventor, even if the AI generated the initial idea. For small molecules, the synthetic challenge is key. Just because AI designs a molecule doesn’t mean it can be made in a lab. The chemist who solves that problem is likely to be seen as the true inventor. However, with large molecules like proteins or nucleic acids—where the AI outputs a sequence that’s straightforward to produce—the line between human and AI contribution becomes blurrier. In such cases, courts may scrutinize whether there was meaningful human input. One major risk for scientists and companies is over-documenting AI’s role. If lab notebooks or publications suggest AI was central to the invention, it could backfire in litigation. Opponents might argue that there was no human inventor, invalidating the patent. Therefore, accurate and careful documentation is crucial—highlighting human judgment, refinement, and decision-making, even when AI is used. AI developers rarely hold patents on AI-assisted inventions unless they played a direct, inventive role—such as fine-tuning a model in a way that led to a breakthrough. In most cases, the pharma company using the AI owns the resulting IP, especially if the contract assigns all rights to them. Service agreements typically give users full ownership of outputs, but companies should ensure contracts explicitly state this to avoid future disputes. Beyond patents, AI companies can still profit through commercial arrangements. Outcome-based pricing—where the AI provider earns more if the drug succeeds—is possible through private contracts. However, such deals depend on negotiation and can be influenced by power imbalances between large AI firms and pharma companies. As AI becomes essential to drug discovery, clear rules and contracts are urgently needed. The faster development timelines investors now demand mean companies can’t afford delays in resolving ownership questions. Setting these frameworks early will not only protect innovation but also accelerate the delivery of new treatments to patients.