The 18-Month Disclosure Window That Hands Chinese AI a Biotech IP Head Start

A 2024 industry survey of US biotech firms found that 79% had at least one contract or product agreement with a China-based or Chinese-owned manufacturer. Sen. Jim Banks is now pointing to a second vulnerability stacked on top of that supply-chain exposure: the intellectual property itself.

In a letter to Commerce Secretary Howard Lutnick, the Indiana Republican asked the US Patent and Trademark Office (USPTO) to keep pursuing reforms that would protect American inventions from what Banks called "AI-enabled duplication." Banks argues that Chinese firms are combining AI tools with traditional patent analysis to scan US biotech applications, identify commercially promising discoveries, and file derivative patents with modifications small enough to win fresh regulatory approval before the original American innovator reaches market.

The mechanism is closer to arbitrage than to plagiarism. A US patent application is published 18 months after filing under standard procedure, long before most biotech drugs clear clinical trials and reach approval. AI models can read the application, identify the molecular target or mechanism of action, and propose minor modifications (salt forms, dosage forms, combination tweaks) that are patent-eligible in their own right. China's drug regulator has moved approvals to historic speed in recent years, which compresses the race between the original US innovator and any domestic competitor that successfully files a derivative claim.

"Patent scraping is intensifying the Chinese IP threat and risks undercutting American R&D investment, particularly given China's rapid drug approval process," Banks wrote.

The mechanism Banks describes is distinct from the supply-chain fight that produced the BIOSECURE Act in 2024. BIOSECURE targeted US dependence on Chinese biomanufacturing, particularly contract development and manufacturing organizations. The patent concern sits on a different layer. The US patent system requires inventors to publish their work in exchange for protection. AI, in Banks' characterization, has turned that disclosure into an arbitrage window.

Banks also makes the structural point that the supply-chain exposure was a US choice. He notes that US drugmakers shifted active pharmaceutical ingredient (API) and generic manufacturing to China for cost reasons over decades, leaving the dependency that the patent layer now sits on top of. That framing turns the USPTO ask into the second layer of a defense an industry with documented supply-chain exposure arguably needs.

The two exposures are connected. An industry whose manufacturing already runs substantially through Chinese facilities is, per Banks' framing, also an industry whose patent disclosures now reach Chinese AI systems first. BIO member companies have warned separately about the supply-chain layer. The trade group has not publicly assessed Banks' IP-scraping characterization, and no independent industry or legal source in the public record quantifies how systematically "AI-enabled duplication" is being deployed.

The reform venue Banks is targeting is not a blank slate. USPTO under Director Squires is already pursuing AI-patent examination reform, work that began before the biotech angle surfaced. Whether those existing tools can address derivative-filing arbitrage, or whether they would need a separate biotech-specific overlay, is unresolved.

Defensive options on the table include tighter examination timing, biotech-specific defensive publication rules, or restrictions on AI use during patent prosecution. Each changes the trade at the patent system's core: how much disclosure an inventor owes in exchange for protection. Banks has placed his biotech letter inside a longer pattern of AI national-security advocacy, which included a May 2026 call for AI oversight. The patent office's response to Lutnick's referral is the next marker to watch.