Using Generative AI in Your Invention Process: What Patent Law Still Requires

Generative AI has made it easier than ever to brainstorm ideas, explore technical solutions, and iterate on product designs. But if you're using AI as part of your inventive process, patent law has not changed to match: a human inventor is still required, and the quality of your own contribution determines whether your invention is protectable.

Patent lawyer Kushal Shah of Perley-Robertson outlines what inventors need to know now that AI-assisted invention is becoming standard practice.

Key Takeaways

• AI can assist, but cannot be the inventor. Patent systems worldwide require a natural person as the inventor. Using AI as a tool is permitted; accepting its output without meaningful human intellectual input is not a solid basis for a patent claim.
• Document everything during the process, not after. If your patent is challenged, you may need to demonstrate the specific contributions you made — which prompts you developed, which AI outputs you accepted or rejected, what modifications you made, and why. Records created in real time are far more persuasive than reconstructed memories.
• Risk of simultaneous similar filings is higher with AI. Because AI tools are trained on overlapping data and may return similar outputs for similar prompts, independent inventors working on the same problem may file competing patents. This makes novelty and non-obviousness harder to establish.

Shah also flags a practical risk many inventors overlook: AI tools can generate technically convincing but flawed or impractical solutions. Before filing a patent application, inventors should independently verify that the proposed solution actually works and that they can explain why — not just that an AI suggested it.

The article also covers confidentiality risks of entering sensitive invention details into AI platforms, and Canadian-specific guidance on current CIPO practice around AI-assisted inventions.

Read the full article on Ottawa Business Journal.