Protecting AI in Biopharma Drug Discovery Against Intellectual Property Theft

Recent discussions in the biopharma sector highlight the urgent need for companies to safeguard their artificial intelligence (AI) datasets and models from intellectual property (IP) theft. As AI becomes increasingly integrated into drug discovery, the potential for misuse of proprietary data—such as clinical trials and genomic information—poses significant risks. Natasha Daughtrey, a partner at Goodwin specializing in IP litigation, emphasizes that without robust protective measures, companies may find their innovations compromised.

The reliance on vast datasets for training AI models in biopharma underscores the importance of stringent contractual agreements. Daughtrey advises that every contract involving proprietary data should clearly define permissible uses, especially regarding model training. This proactive approach is essential as AI models can inadvertently incorporate proprietary information, making it challenging to untangle ownership and rights once misappropriation occurs. The legal landscape is still evolving, with courts grappling with how to address cases where proprietary prompts or other elements of AI systems are allegedly stolen. Companies must treat their training processes as trade secrets, implementing rigorous access controls similar to those applied to their most sensitive information.

Strategically, biopharma firms must navigate the complex decision of whether to protect their innovations through patents or trade secrets. Daughtrey suggests that the choice should be made on a case-by-case basis, considering factors such as patentability and the nature of the innovation. If a competitor can easily reverse-engineer an innovation from a product, a patent may be more beneficial. Conversely, if the value lies in aspects that are not readily discernible, such as proprietary datasets or internal processes, trade secret protection may be more appropriate. This nuanced approach allows companies to maximize their competitive advantage while minimizing the risk of disclosure.

Litigation risks are also on the rise, particularly regarding inventorship disputes linked to AI contributions. The United States Patent and Trademark Office (USPTO) has clarified that AI cannot be named as an inventor, reinforcing the need for human inventors to demonstrate their contributions to the inventive process. This requirement presents challenges, as companies must maintain thorough documentation of the role that AI played in their research. As AI tools become more prevalent, the legal landscape will likely see an increase in disputes over patent validity, particularly as defendants scrutinize the extent of AI's influence on claimed inventions.

The implications for the biopharma industry are significant. As AI tools become commonplace, the bar for establishing patent obviousness may lower, complicating the patenting process for innovative drugs. Conversely, the ease of experimentation facilitated by AI might enhance the ability of patentees to demonstrate enablement, potentially strengthening their claims. Companies must adapt their patent strategies to account for these shifts, ensuring that they can effectively defend their innovations in an increasingly competitive and litigious environment.

Moving forward, biopharma firms should prioritize the integration of comprehensive inventorship documentation into their research workflows. Establishing clear records of human contributions in the context of AI-assisted inventions will be crucial for defending against future patent challenges. As the industry continues to evolve, those that proactively address these legal complexities will be better positioned to capitalize on the transformative potential of AI in drug discovery while safeguarding their intellectual property.