Background
Artificial intelligence (AI) is transforming the life sciences industry, with potentially its most significant impact on drug discovery and development including the acceleration of inherently burdensome tasks such as target identification, lead optimization, and safety and efficacy prediction. [1]
AI models trained on multiomics, literature, clinical phenotypes, and patient and other data can identify candidate disease biology hypotheses and entirely new protein and pathway targets. [2] Some systems can even infer receptor–ligand interactions from correlated gene expression signatures, without prior structural annotation. [3]
Generative AI, such as graph neural networks, diffusion models, and molecular design systems, has also been reported to propose novel chemical entities predicted to bind to selected targets. [4] Since composition-of-matter patents can protect entire therapeutic platforms, these new compound outputs may represent the most commercially valuable results of an AI platform.
Numerous companies have emerged offering AI drug discovery platforms to either serve others in drug development and/or advance their own pipelines. [5] These approaches raise acute challenges in patentability (inventorship, obviousness, enablement) as well as management of data and confidential information. Counsel must therefore carefully account for risks and opportunities unique to AI-enabled drug discovery.
Hypothetical Deal Scenario
For the purposes of this article, we consider the following deal example. Company A, an AI-platform company, has created a compound library through its platform and seeks to license its library and platform to others. Pharma Company B wishes to license selected compounds from the library and use the platform to discover new compounds for an oncology indication (“Indication X”).
Company A’s interests include maximizing the value of its compounds and services revenue while protecting its background IP. Company B’s interests include obtaining ownership rights over the IP relating to the compounds used for Indication X.
Library Compounds
The terms for licensing the library compounds would include typical provisions like upfront fees, milestones, and royalties on sales of products covered by applicable patents. Company B will likely insist on an exclusive license to the library compounds, while Company A may limit exclusivity to a defined field (e.g., a specific target or indication) to preserve rights elsewhere, allowing Company A to segment and monetize the same compounds across multiple therapeutic areas.
Platform-Generated Compounds
For compounds discovered via the platform, Company B may seek outright ownership, compensating Company A through service fees.
Alternatively, if the resulting compound class has broader utility in indication areas distinct from Indication X, Company A may seek to retain ownership but grant Company B exclusivity only for Indication X. If this concept is acceptable to Company B, the compounds should be treated as part of Company A’s library and licensed accordingly with lower service fees.
Joint ownership rights in the compounds is also an option, but that could complicate matters including with regard to attracting investment and risking that full commercialization rights do not vest with a single party.
Inventorship Considerations
In the United States, only a natural person can be an inventor, [6] and at least one human must make a “significant contribution” to conception of each patent claim. [7]
Courts have yet to define how the “significant contribution” test applies to AI-assisted inventions. Patents issued without proper inventorship – where no human made such a contribution – may be invalidated.
Company B should take this into account when licensing IP developed with the assistance of AI, and should perhaps seek representations and warranties from Company A that inventors named on licensed patents are properly named and that a human made a contribution to the conception of each patent claim. Clauses that terminate royalty obligations upon a finding of invalidity of licensed patents will also help mitigate this category of exposure.
Ownership and Use of Data
Company A would own its AI platform and improvements, but agreements must clearly allocate ownership of data used or generated by the platform, including predictive binding data and toxicity models required by the FDA. [8]
Because AI tools may also be required post-approval for diagnostics, Company B, as the commercialization entity, should ensure perpetual, irrevocable rights to access and use data necessary for such regulatory and commercialization purposes, including the continuing transfer of and access to the data during and surviving the agreement.
Ownership and Confidentiality of Training Data
The parties should specify who owns and may access the training data used by the platform. If Company B contributes such data, it should seek to ensure confidentiality is maintained despite the model’s learning processes.
This can be complicated given that Company A’s platform will doubtless learn from and retain copies of the data. Company B may seek to require that Company A (1) not have access to the raw training data, (2) keep training data confidential indefinitely, (3) not re-use the data for other clients, (4) not copy or reverse engineer the data; and (5) comply with other restrictions.
Given that regulators may review AI-derived data, the platform’s operation should comply with FDA guidance on transparency, validation, and reproducibility. Contracts should also delineate privacy obligations and data-handling standards consistent with HIPAA.
Indemnification and Risk Allocation
AI collaborations introduce new risk vectors. Company B will typically require indemnification from Company A for claims arising from unauthorized use of third-party or restricted data that Company A provides to train the platform. Reciprocal indemnification obligations may apply where Company B supplies the data.
Company B may also seek protection from losses caused by malicious code, technical failure, or cybersecurity breaches and require minimum technical safeguards – encryption, penetration testing, on-prem storage, and disaster recovery – with audit rights to verify compliance.
The parties will likely seek standard indemnification for claims of third-party infringement and require reps and warranties that there is none. Given the potential for significant exposure in the event of a third-party infringement claim, whether and to what extent each party agrees to these clauses will likely be a key point of negotiation.
Patent Prosecution for New Compounds
Patent prosecution rights generally follow ownership. If Company B owns the new compounds, it would control preparation and prosecution, subject to Company A’s review to prevent disclosure of background IP. If Company A owns, it would lead prosecution while coordinating with Company B to ensure alignment on claim scope and geography. Cooperative prosecution models similar to research collaborations can balance interests while confirming inventorship in AI-assisted inventions.
Post-Termination and Bankruptcy Considerations
Agreements should also govern handling of trained models and outputs following term and termination. For example, if Company A customizes an AI model for Company B, Company B could require that the uniquely trained model and data be (1) segregated from Company A’s general systems so as not to be used for the benefit of third parties and (2) after the agreement, be destroyed or otherwise sequestered for Company B’s future use only.
Given that many AI-platform companies are early-stage, Company B should also plan for the potential insolvency of Company A. If regulatory approval or commercialization depends on ongoing platform access, the agreement should ensure continuity even after termination or bankruptcy.
An escrow arrangement could require Company A to deposit a functioning, regularly updated copy of the platform with a neutral agent, enabling Company B to access it for necessary post-approval or other ongoing use. Negotiation may be difficult given, for example, trade-secret concerns, but Company B’s commercial program may depend on this safeguard.
Conclusion
AI-enabled drug discovery raises novel challenges at the intersection of licensing, patent law, and data governance. Licensing counsel must tailor agreements to address inventorship, ownership of compounds and data, confidentiality of training datasets, indemnification, platform continuity, and regulatory compliance. By anticipating these issues, parties can better align innovation incentives and mitigate risk in this rapidly evolving domain.
This article was originally published on the Licensing Executive Society (LES) blog on January 19, 2026. Read more at: https://www.lesusacanada.org/key-ip-licensing-considerations-for-ai-drug-discovery-platforms/.
References
[2] Ocana, Alberto, et al. “Integrating artificial intelligence in drug discovery and early drug development: a transformative approach.” Biomarker Research 13.1 (2025): 45; Wang, Joanna. “Navigating the USPTO’s AI inventorship guidance in AI-driven drug discovery.” Journal of law and the biosciences vol. 12,2 lsaf014. 2 Aug. 2025, doi:10.1093/jlb/lsaf014; see also https://www.drugtargetreview.com/news/166597/ai-platform-detects-new-drug-targets-in-minutes/
[3] Vahid, Milad R., et al. “DiSiR: fast and robust method to identify ligand–receptor interactions at subunit level from single-cell RNA-sequencing data.” NAR Genomics and Bioinformatics 5.1 (2023): lqad030.
[4] Adegbola, Itunuoluwa, Integrating Generative AI with High-Throughput Screening for Accelerated Drug Discovery (June 20, 2025). Available at SSRN: https://ssrn.com/abstract=5317101; Boston Consulting Group & Wellcome Trust, Unlocking the potential of AI in Drug Discovery, (2023) https://wellcome.org/sites/default/files/2023-06/unlocking-the-potential-of-AI-in-drug-discovery_report.pdf
[5] For example, systems like Insilico’s Chemistry42 generate molecules which are then synthesized and tested in vitro. Ivanenkov, Yan A., et al. “Chemistry42: an AI-driven platform for molecular design and optimization.” Journal of chemical information and modeling 63.3 (2023): 695-701.
[6] Thaler v. Vidal, 43 F.4th 1207, 1213 (Fed. Cir. 2022), cert denied, 143 S. Ct. 1783 (2023).
[7] https://www.jdsupra.com/legalnews/key-inventorship-considerations-in-ai-6047492/; “Inventorship Guidance for AI-assisted Inventions,” PTO-P-2023-0043, available at https://www.federalregister.gov/documents/2024/02/13/2024-02623/inventorship-guidance-for-ai-assisted-inventions.
