I am leaving this AI documentation alone, because it’s objective and does not need my opinions. I do want to say one thing, though: Imagine if the patent office allowed AI companies to generate patents. My sense is it would act like CryptoMining with the claims becoming harder and harder to “mine.” So, for today, I think the USPTO (the patent office) has made a cohesive and reasonable set of policies.
Here’s what the USPTO has formally said and done on AI and intellectual property, with the most current positions:
- Human inventorship is required
- Only natural persons can be named inventors on U.S. patents (reaffirming Thaler v. Vidal). AI systems cannot be inventors, though AI tools may assist in the inventive process.uspto+1?
- AI-assisted inventions are not categorically unpatentable. Patentability turns on whether one or more humans made a “significant” contribution to the claimed invention; examiners will look for documented human contribution commensurate with the claims.federalregister+2?
- 2024 Inventorship Guidance (in force)
- Effective Feb 13, 2024, the USPTO issued Inventorship Guidance for AI-assisted inventions (Federal Register). It instructs examiners and applicants on evaluating human contribution in AI-assisted applications and provides examples; public comments were invited through May 13, 2024.uspto+1?
- Key takeaway: show how specific claim elements reflect human ingenuity; improper inventorship can risk validity, so applicants should carefully attribute and evidence human contributions.ropesgray+1?
- Subject-matter eligibility and examination direction
- The USPTO has signaled examiner-facing updates and memos affecting AI/software eligibility and examination, including efforts to clarify treatment of AI-related claims and refocus standards; practitioner alerts in 2024–2025 describe a patentee-friendly shift and later memos aimed at tech centers handling AI/software.alston+2?
- Agency AI Strategy (2025)
- In January 2025, the USPTO announced a new AI Strategy guiding policy and operations: develop IP policies that promote inclusive AI innovation, build AI capabilities and infrastructure, promote responsible AI use at USPTO, develop internal AI expertise, and collaborate across government and internationally. It aims to foster responsible AI innovation and harness AI to improve patent/trademark examination (e.g., prior-art search).uspto+2?
- Pilot programs and tools: the Office has been integrating AI in search/exam and launched initiatives like the AI Search Automated Pilot (ASAP!) in 2025 to enhance examination efficiency.uspto+1?
- Broader policy development
- Since 2019–2020 the USPTO has collected stakeholder views and published reports on AI and IP policy across patents, trademarks, copyright, trade secrets, and databases, committing to evolve guidance alongside technological change.uspto+1?
What this means in practice for applicants and owners
- Document human contributions: Clearly map human inventive acts to claimed elements; maintain detailed records of conception and reduction to practice where AI tools were used.federalregister+2?
- Prosecution strategy: Expect examiner scrutiny on inventorship and disclosure around AI use; leverage USPTO examples/guidance to frame claims and specifications.uspto+2?
- Watch eligibility updates: AI/software claims may benefit from evolving examiner guidance and memos—track subject-matter eligibility developments and appeal decisions cited in USPTO communications.slwip+2?
- Operationally, anticipate more AI-enabled search/exam tools at the Office that can affect prior-art identification and distinctiveness analyses.uspto+2?
Note on copyright (separate agency): The U.S. Copyright Office (not USPTO) has been issuing AI reports (2025) on training data, fair use, and outputs; these shape the copyright side of AI but are distinct from USPTO patent policy.
Edited by
Erik Linask
